Similar study for passenger cars was conducted under MOU 39

1
Effects of ACC heavy-duty vehicles in mixed traf
fic environmental effects

by Margareta Stefanovic and Petros Ioannou Cente
r for Advanced Transportation Technologies
University of Southern California
Los Angeles, CA 90089 http//www.usc.edu/dept/ee/
catt
2
OUTLINE OF PRESENTATION

• Motivation
• Truck model
• ACC control law
• Human Driver Model
• Vehicle Following in Mixed Manual/ACC Passenger
  Cars/HDV Traffic
  
• Lane Change Effects
• Conclusions

3

• Motivation
• Several controllers for speed tracking and
  vehicle following for HDVs have been designed
  and tested recently (e.g. by UCLA research
  group). They demonstrated local and string
  stability and close vehicle following even in the
  case of large actuator delays.
• The controllers were tested for performance in
  HDV platoons, designed to operate in dedicated
  lanes. Of interest here is the impact of the
  mixed manual/ACC passenger car/HDV traffic on
  traffic flow characteristics.
• What is the effect of ACC HDVs on environmental
  pollution and fuel efficiency? Similar study for
  passenger cars was conducted under MOU 392 and TO
  4217.
• Large gap formed in front of the truck (both ACC
  and manual) in high acceleration maneuvers gives
  rise to cut-ins in front of the truck and
  potentially dangerous situations. What is the
  impact on traffic flow characteristics and
  emission when trucks are ACC equipped?
• D.Yanakiev, J. Eyre and I. Kanellakopoulos
  Analysis, Design, and Evaluation of AVCS for
  Heavy-Duty Vehicles with Actuator Delays Final
  Report for MOU 240, California PATH Research
  Report UCB-ITS_PRR-98-18

4
Truck model
Complex nonlinear model of the longitudinal truck
dynamics was compiled by the UCLA group. After
linearization and neglecting of fast modes a
simple 1st order linear model is obtained
- truck longitudinal speed
- fuel / brake control input
d - disturbance
parameters a, b depend on the operating point
(steady state speed and the load torque) around
which the system dynamics is linearized.
5
- Based on the linearized model, various
longitudinal controllers for vehicle following
had been developed and tested for stability and
performance.
- Among several such controllers we select a
simple fixed gain PID controller, as it was shown
that, with the appropriate choice of control
parameters and spacing policies, stability and
good performance of vehicle following are
achieved, while the controller complexity is
greatly reduced..
Control objective
Regulate to zero relative velocity vr and
separation error ?
- desired inter-vehicle spacing
This is achieved by satisfying
(k - design parameter (constant or varying)
tuned according to the performance requirements.
6
ACC controller
-We select a fixed gain PID control law with the
approximate derivative term
Spacing policies
- variable time headway
if

introduced to reduce the steady state
inter-vehicle spacing and to smooth the control
action, and

- variable position error gain
included to limit the acceleration/deceleration
that results from the large spacing error. Note
that exponential term in k has a quadratic
dependence on the position error d.
7
Truck Human Driver Model

• There exists a variety of human driver vehicle
  following models for the passenger vehicles.
  
• Very few of them are designed for heavy
  vehicles.
  
• The Bando vehicle following model

V(?x) is the desired safe velocity, determined by
the following distance from the preceding
vehicle, and has a sigmoid shape


in this model is modified to be used for
trucks in the following sense lower sensitivity
of the HDD vehicle due to its lower
actuation-to-weight ratio is accounted for in the
smaller sensitivity constant a (typically 0.8 or
less vs. 1.5 5 for cars).

M. Bando et al Dynamical Model of Traffic
Congestion and numerical simulation, Physical
Review E, Vol. 51, N0 2, February 1995
A.Mason and A.Woods Car-Following Model of
Multi-species Systems of Road Traffic, Physical
Review E, Vol. 55, N0 3, March 1997.
8
- We modify Bando model further, to account for
the fact that the truck driver can see further
away when he follows the passenger cars ahead of
him Instead of depending on the following dis
tance from the immediately preceding vehicle
only, the desired safe velocity is assumed to
depend on two (or more) vehicles ahead, e.g.


The proportionality constant ? (0 chosen so that the sensitivity with regards to
the immediately preceding vehicle has the highest
weight (? 0.5 in the above equation).

9
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic

Consider a fleet of ten vehicles in a single
lane, nine of which are manual passenger cars,
and one is ACC truck. Proposed parameters from
UCLA group ck 0.1, k0 1, s 0.1, H0
0.1, ch0.2 Kp 150, Ki3, Kd20 Our tuned pa
rameters ck 0.01, k0 0.2, s 1, H0 0.4,
ch0.2 Kp110, Ki1, Kd100 Exponential term
in position error gain is chosen to have a
parabolical dependence on the position error, as
proposed in . For low accelerations of the prec
eding vehicles, the actuator of the ACC truck
does not saturate, and the control action that is
commanded results in a stable truck behavior,
with close vehicle following, as seen from the
figures below D.Yanakiev, J. Eyre and I. Kanel
lakopoulos Analysis, Design, and Evaluation of
AVCS for Heavy-Duty Vehicles with Actuator
Delays Final Report for MOU 240, California
PATH Research Report UCB-ITS_PRR-98-18


10
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic

Speed responses for the mixed manual/ACC traffic
(4th vehicle is ACC truck, the
rest are manual passenger cars)


Position error vs. time for the ACC truck
11
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
Speed responses for the 100 manual traffic (4th
vehicle is manually driven truck)
12
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic

- This choice of functional dependence of
position error gain k on the position error is
justified for relatively low acceleration rates
of the lead vehicle, as well as for the case of
truck platooning.

• Here we do not consider platooning of HDVs, but
  the mixing of passenger cars and trucks on
  highways, when vehicles in front may attempt to
  accelerate at the rates higher that those
  attainable by the heavy vehicle dynamics.

13
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
Speed responses for the mixed manual/ACC traffic
(4th vehicle is ACC truck, the
rest are manual passenger cars). Parameter k
depends on position error squared.
14
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
where the position error of the ACC truck is
Position error vs. time for the ACC truck
Note the collision between the truck and the
passenger car in front.
For the purposes of vehicle following in high
acceleration scenarios, we modify the expression
for the position error gain to have the
exponential term linearly dependent on the
position error.
15
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
With this modification, the above scenario is
simulated again
Clearly, the ACC truck keeps a safe distance from
the vehicle ahead at all times.
16
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic

The performance of the 100 manual traffic in the
same scenario is now simulated. Bando model for
the truck driver is used, with the safe desired
speed dependent on the two vehicles immediately
ahead
17
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
- Environmental Evaluation

• Compare the accumulative pollution levels and
  fuel economy of the 100 manual traffic
  (passenger cars and trucks) vs. mixed traffic
  (manual cars and ACC trucks).
• Environmental evaluation model used For
  passenger cars CMEM - Comprehensive Modal
  Emissions Model (UC Riverside, 1998) high
  fidelity model, sensitive to transients. For
  trucks Heavy-Duty Diesel Modal Emissions and
  Fuel Consumption Model (being developed by UC
  Riverside group, TO 4215).
• Tailpipe emissions of unburnt hydrocarbons HC,
  carbon monoxide (CO), oxides of nitrogen (NOx),
  CO2 and fuel consumption are calculated and
  compared for mixed vs. manual traffic.

18
Vehicle Following in Mixed Manual/ACC Passenger
Cars/HDV Traffic
For the last shown speed scenario in a single
lane vehicle following, the calculated benefits
are given below
However, due to the low acceleration capabilities
of the HDVs, there is a very large position error
formed in both mixed and manual traffic case. In
multilane highway situations, this will
undoubtedly give rise to (multiple) lane cut-ins
in front of the truck.
19
Lane Change Effects
Consider the same speed scenario now vehicles
from the adjacent lanes cut in front of the
truck, using the gap formed in front of it
Speed responses in mixed traffic 4th vehicle is
ACC truck.
Position error vs. time of the ACC truck.
20
Lane Change Effects
- When ACC truck is replaced with the manually
driven one, less cut-ins are expected in front of
it, due to the less smooth speed response of the
truck human driver. We assume one cut-in from the
neighboring lane
Speed responses in 100 manual traffic 4th
vehicle is manual truck.
21
Lane Change Effects
The environmental effects?
Due to more traffic flow disturbances in the
mixed case (two cut-ins instead of one) the
benefits are correspondingly reduced
22
Lane Change Effects
- How are the benefits affected when the cut-in
vehicle is an ACC equipped passenger car?
Speed responses in mixed traffic 4th vehicle is
ACC truck, cut-in vehicle is an ACC passenger car.
- Due to smooth acceleration of the cut-in
passenger car, we assume less cut-ins in front of
ACC truck.
23
Lane Change Effects
Calculated environmental effects
24
Conclusions

• ACC equipped trucks mixed with passenger cars in
  a vehicle following with no lane change result in
  moderately improved traffic flow characteristics
  as well as emission levels and fuel efficiency.
• Due to low acceleration capabilities of HD
  vehicles, frequent cut-ins in front of the trucks
  are expected in high acceleration maneuvers
  performed downstream. This is particularly so
  when the truck is ACC equipped.
• Lane cut-ins result in reduced benefits on the
  mixed traffic side.
  
• A conclusion to be verified is that higher
  benefits are obtainable if trucks are platooned
  instead of mixed individually with other vehicles
  on the road.
• Further investigation and design of new ACC
  control laws to achieve higher benefits for mixed
  manual/ACC truck/passenger cars traffic is
  underway.