Title: Microscopic Traffic Flow Models
1Microscopic Traffic Flow Models
- Its probabilistic nature and its consequences for
subsequent congestion
Physics of Traffic
Qing Ou email Q.Ou_at_tudelft.nl
2Topics
- The earlier microscopic models (CA, CF)
- KKW model
3Cellular Automaton (CA) Model
- Space is divided into cells
- Time is discrete
- Rules
- CA Models are used to described discrete
dynamical systems
41D CA Model for Microscopic Traffic
- Nagel Schreckenberg (NS) Model
- A road is divided into L cells numbered by 1,2L
- Time is discrete
- Vehicle with integer velocity v0,1,Vmax
5Rules for NS Model
- Acceleration viltVmax, then vi? vi 1
- Slowing down if vigtdi then vidi, (dixi1-xi-1
and xi is the position of vehicle i ) - Stochastic deceleration with probability p,
vi(gt0)? vi -1 - Position updating xi? xi vi
- (Introduction of probability make it difficult
to get analytic solution to this model)
6Fukui Ishibashi (FI) Model
- Velocity updating vi?min(di, Vmax)
- Position updating xi? xi vi
- An analytical relation between average velocity
and average density -
7Car Following (CF) Model
- Three main types
- GHR model
- Safety-distance model (Gipps model)
- Linear model
- Psycho-physical (e.g. GM models)
- Reactionfollowersensitivityfollowerstimulus
8Car Following (CF) Model
- The basic ideas about car following model can be
summarized - The function represents the stimulus to the
vehicle, and this stimulus is composed by the
speed of the vehicle speed difference and gap
between the vehicle and the leading vehicle
9Gazis-Herman-Rothery (GHR) model
- GHR Model well-known in late fifties and sixties
- which considers relative spacing and speeds
between vehicle n and n-1. - c, l, m are constants and T is driving
reaction time
10Best combination of m and l regarding GHR model
- m0.8, l2.8 (May and Keller, 1967)
- m-0.8, l1.2 (Heyes and Ashworth, 1972) data
from Mersey tunnel in UK - m0.6 , l2.4 (Ceder and May, 1976) a far large
number of data sets - replaced by , S is jam
spacing and A valued from 0 to 10 indicating free
flow to congestion - Now seldom used for large number of contradictory
findings as to correct m and l.
11Car Following (CF) Model
- 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
12Linear Model
- Acceleration
- Desired following distance
-
- Share the similar disadvantages with GHR model
13Two Lane Model (Symmetry)
- The single lane model results in platooning with
slow vehicles followed by faster ones. - The most important elements of the two lane
model - Symmetry
- Stochasticity
- Direction of Causality
14Basic rules for lane changing
- Look ahead if somebody is in your way
- Look on the other lane if it is any better there
- Look back on the other lane if you would get in
somebody elses way
15Technical rule for lane changing
- T1 how far you look ahead on
your lane - T2 ahead on the other lane
- T3 back on the other
lane - T4
- When all conditions are satisfied, lane changing
16The main parts of complete microscopic model
- Motion rules
- Lane changing rules
- Stochastic ingredients
17Microscopic three-phase traffic theory (KKW Model)
- Earlier traffic flow theories and models are in a
serious conflict with many of these empirical
spatiotemporal traffic pattern features -
- Introduction of three-phase traffic theory to
explain all eimpirical spatiotemporal congested
pattern. - Free flow
- Synchronized flow
- Wide moving jam
18Main rules in KKW Model
- Vehicle Motion rules
- Lane changing rules
- Random acceleration and deceleration rules
- The biggest feature in KKW model
- Synchronization distance (page 100)
- (E q16.29
page410)
19Main rules in KKW Model
20Main rules in KKW Model
- Lane changing rules
- Incentive conditions (Eq16.75 p420)speed
- Security conditions (Eq16.77 p421)gap
- depending on the function
21Main rules in KKW Model
- Stochastic part of KKW model
22Main behavioral model assumptions and model
parameters
- In synchronized flow, a driver accepts a range of
different hypothetical steady state speeds at the
same space gap to the preceding vehicle.
23Main behavioral model assumptions and model
parameters
- A driver tends to adjust the speed to the
preceding vehicle within the synchronization
distance
24Main behavioral model assumptions and model
parameters
- Over-acceleration effect
- It is responsible for an F?S transition and the
related Z-shaped dependence of vehicle speed on
density - The simulation of the vehicle over-acceleration
effect is made through the use of random vehicle
fluctuations - Over-Deceleration effect
- This random effect of the vehicle
over-deceleration is responsible for moving jam
emergence. - (Stochastic part of model, page 426)
25Main behavioral model assumptions and model
parameters
- A driver in synchronized flow does not accelerate
before the preceding vehicle has begun to
accelerate. - Moving in synchronized flow, a driver comes
closer to the preceding vehicle over time that
explains the pinch effect in synchronized flow. - (Stochastic time delay of acceleration and
deceleration, page 426)
26Conclusion
- Three-phase theory can explain empirical features
of phase transitions and congested patterns at
freeway bottlenecks. - Models based on this theory is formed by the
introduction of a synchronization distance - The synchronization distance depends on
time-dependent vehicle speeds. - Safety conditions, driver time delays, stochastic
behavior, lane changing rules should be well
adjusted
27