A Glimpse inside the Black Box Network Microsimulation Models

1
A Glimpse inside the Black Box Network
Micro-simulation Models

• Ronghui Liu
• Institute for Transport Studies
• University of Leeds, UK
• r.liu_at_leeds.ac.uk

2
Contents

• Introduction
• Individual model components
• Practical implementations
• Summary

3
Introduction to micro-simulation

• A numerical technique for conducting experiments
  on a digital computer, which may involve
  mathematical models that describe the behaviour
  of a transportation system over extended periods
  of time. (May 1990)

4
Keywords

• System the real-world process to imitate
• Model the set of assumptions, in the form of
  mathematical or logical relationships, put
  forward to help understand how the corresponding
  system behaves
• Entities people or vehicles that act in the
  system
• Time an explicit element of the system.

5
The transport road network systems

• Demand for travel activity-based trip chains and
  trip timing
• Route choice (pre-trip and en-route)
• Departure-time choice
• Traffic movements
• Public transport (bus) systems

6
Interaction of the model systems
7
Micro-simulation models of route choice

• Bounded rational model
• Stay on the same habit route unless the
  alternative is well better (Mahmassani et al)
• Probabilistic discrete choice
• From individual perceived costs on alternative
  routes
• Myopic switch
• Always takes the minimum cost route

8
Models of departure-time choice

• Preferred arrival time (Small 1982)
• Probability distributions of inter-departure
  headways based on average flows of a given demand
  profile

9
Traffic micro-simulation entities

• Driver and vehicle characteristics.
• Physical size length and width
• Mechanical capacity maximum acceleration or
  deceleration
• Driving behaviour desired speed, reaction time,
  gap acceptance, aggressiveness, etc.

10
Traffic micro-simulation models

• Car-following model
• Lane-changing model
• Gap-acceptance model
• Lane-choice model
• Models of intersection controls

11
Car-following models

• Models of individual vehicle following behaviour
• In a single stream of traffic (lane disciplined)
• No overtaking
• Three main types
• Safety-distance model
• Action-points different rules for different
  driving states
• Psycho-physical
• AccelerationStimulus x Sensitivity

12
Model requirements

• Agree with experimental evidence
• Microscopic individual vehicle trajectories
• Macroscopic q-k-u relationships
• Be psycho-physically feasible
• Posses local stability
• Perturbations in behaviour of lead vehicle not
  causing following vehicle to collide
• Possess asymptotic stability
• Perturbations not magnified back over a line of
  vehicles

13
The Gipps car-following model

• Free flow model
• Accelerate freely to desired speed
• Safety-distance model
• Driver maintains a speed v which will just allow
  him to stop in emergency without hitting the
  obstacle at distance S ahead

14
Variants and constraints

• Variable reaction times
• Variable acceleration and deceleration
• Variable or multiple lead vehicles
• Lane-disciplined
• Stable traffic flow do not produce incidents

15
Gap-acceptance models

• Models of individual drivers choice of safety
  gaps to merge into or to cross other traffic
  streams
• Two elements
• Gaps acceptable to drivers
• Gaps available to the driver

16
Variants and constraints

• Time-dependent acceptable gap
• Courtesy yielding
• Individual gap acceptance no shadowing effects
  (e.g. on approaching roundabouts)
• Requires distinction of major/minor flows

17
Lane-changing models

• Models of individual drivers ability and
  propensity to change lanes
• Lane-changing objectives, e.g.
• To overtake a slower moving vehicle
• To bypass an obstacle
• To move off/into a reserved bus lane
• To get-in-lane for next junction turning
• To giveway to merging traffic
• Decision-making behaviour
• Is it possible to change lane? (physically
  safely)
• Is it necessary to change lane? (for junction
  turning?)
• Is it desirable to change lane? (to overtake?)

18
Variants and constraints

• Variable lane-changing objectives
• Variable hierarchical decision trees
• Variable acceptable gaps
• Look-ahead anticipating a lane-changing needs a
  link ahead
• Cooperative lane-changing
• Courtesy yielding
• Lane disciplined no overtaking in between lanes
  or lane in opposite direction

19
Lane-choice models

• Selection of the lateral position in entering or
  traversing a link.
• Pre-specified, or
• Instantaneous choice made in response to traffic
  condition and destination

20
Models of intersection traffic control

• Signal controlled fixed or responsive
• Priority giveway
• Roundabout partially signalised
• Motorway merge
• Stop-and-go
• Giveway to oncoming traffic

21
Network representation

• Network topology
• Junction type and priority rules
• Link (major/minor, speed limits, ..)
• Lanes (turning restriction, access restriction)
• Signal plans (stages, phases, responsive rules)

22
Software implementation

• Discrete time vs. discrete event
• Fixed time increment used to simulate systems
  whose entities change continuously with time,
  e.g. traffic flow, the speed of vehicles
• Event scanning the system is updated every time
  a main event takes place. Used to model systems
  whose entities change instantaneously at separate
  points in time, e.g. traffic signals

23
Software implementation (II)

• Continuous space vs. cellular automata (CA)
• CA represents road sections as fixed-length
  segments and vehicles jump from one segment to
  another
• fast, easy to implement on parallel machines but
• may induce large speed changes

24
Practical micro-simulation approach

• Pure traffic micro-simulation
• Routes drawn from turning probability
• May lead to implausible cyclic route
• Traffic micro-simulation with no or simple route
  choice
• Route choice from static equilibrium model, or
  based on aggregated feedback loop from the
  micro-simulation
• Lack of consistency in route choice mechanism
• Day-to-day micro-simulation of route and
  departure-time choice
• Based on simple traffic model u-k relationships
• Lack of junction modelling
• Difficult to deal with mixed traffic, bus lane,
  traffic controls at intersections

25
The DRACULA approach

• Dynamic Route Assignment Combining User Learing
  and microsimulAtion
• A micro-simulation model of the full supply
  systems
• Micro-simulation of day-to-day route and
  departure-time choices
• Micro-simulation of traffic movements
• Micro-simulation of public transport systems and
  passenger route choice

26
Summary

• Traffic micro-simulation models deal naturally
  with time-dependent queues, lane sharing
  problems, variability, etc
• They are based on simple mathematical models and
  logical rules.
• There are varied implementations in networks.
• Greater efforts required to adopt it as a
  suitable traffic model for DTA