Mainstream traffic flow control at sags
We present a new control strategy that aims to reduce total delay at sags. The strategy is based on the concept of mainstream traffic flow control. The traffic density at the bottleneck area is regulated in order to keep it slightly below the critical density, hence preventing traffic from breaking down while maximizing outflow. Density is regulated by means of a variable speed limit section that regulates the inflow to the bottleneck. Speed limits are set based on a proportional feedback control law. We evaluate the effec-tiveness of the control strategy by means of a case study using microscopic traffic simulation. The results show a significant increase in bottleneck outflow, particularly during periods of high demand, which considerably reduces delay.
Abstract
Bernat
Goñi Ros
, Victor L.
Knoop
, Bart
van Arem
, Serge P.
Hoogendoorn
Sags are bottlenecks in freeway networks
cffree csfree Density (veh/km) Flow (veh/h) cfcong cscong Flat section Sag bottleneck Increase in gradient (sag) Increase in resistance force Insufficient throttle operation
Vehicle acceleration limitation
Local changes in car-following behavior: - Lower free flow speeds
- Longer headways at a given speed
Reduction in traffic flow capacity (10-20%)
Insufficient propulsion force
Control strategy
d s Bottleneck VSL section Controller Speed limit (vlim) Density measurements (ρ
) qcRegulate the inflow to the bottleneck ( cscong < qc < csfree ), so that the bottleneck does not activate:
• Exit flow (s) = VSL section outflow (qc) > discharge capacity (cscong)
• VSL section outflow (qc) < demand (d)
Transportation Research Board
93rd Annual Meeting
January 12-16, 2014
Paper 14-2726
Bernat Goñi Ros
Delft University of Technology
TRAIL Research School
b.goniros@tudelft.nl
Conclusions
Case study setup
Results
Target speed Proportional feedback:v
lim(k) = v
*lim+ K
p· [
ρ
*−
ρ
(k
−
r)
]
Gain Target density Delay
v
lim(k) = [v
lim_min, v
lim_max]
Speed limit constraints
v
lim(k)
is a multiple of 10|
v
lim(k
+
1)
−
v
lim(k)
| ≤
∆
v
lim• The controller reacts adequately to changes in demand
• 7% increase in exit flow (s) in periods of high demand
• 30% reduction in total delay / Significant reduction also
with different car-following model parameter values
• Traffic composition: 100% passenger cars, 1 driver class
• Car-following model:
Acceleration = f ( speed, relative speed, spacing, gradient)
100% compliance with speed limits
• Scenarios: a) Reference scenario (no influence of gradient)
b) No-control scenario c) Control scenario
• Network performance measure:
Total delay (TD) = Total travel time (TTT)
−
TTTReference• Simulation time: 10000 s
• Network: 30 km, 1 lane, 1 sag
• The proposed control strategy has the potential to reduce
total delay at sags
• Further research:
- Evaluation: multi-lane network, heterogeneous traffic
- Controller design: mitigation of oscillatory behavior,
combination with other control measures
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0 500 1000 1500 2000 Demand flow
Exit flow (Reference scenario) Exit flow (No−control scenario) Exit flow (Control scenario)
0 2000 4000 6000 0 500 1000 1500 F lo w ( v e h /h ) Simulation time (s) 8000 10000 2000 x = 0.3 km x = 29.9 km 0 20 40 60 80 100 120
Traffic speed on the VSL section
Speed limit displayed on the VSL section
0 2000 4000 6000 20 0 40 60 80 100 120 S p e e d ( k m /h ) Simulation time (s) 8000 10000 x = 27.0 km 25000 26000 27000 28000 29000 30000 0 20 40 60 80 100 120 25.0 26.0 27.0 28.0 29.0 30.0 20 0 40 60 80 100 120 A lt it u d e ( m ) Location x (km) Constant slope –0.5% Constant slope +2.5% VSL section Detector at the bottleneck 60 60 60
With high demand (d > csfree), the bottleneck activates:
• Exit flow (s) = discharge capacity (cscong) < demand (d)
d s Bottleneck 500-1000 m d s Bottleneck
With low demand (d < csfree), the bottleneck does not activate:
• Exit flow (s) = demand (d)
