Traffic Flow

Fundamentals
CE331 Transportation Engineering
Objectives

 Understand the fundamental

relationships among traffic parameters
 Estimating traffic parameters using
the fundamental relationship
 Queuing Models

2
Some Terms

 Speed (u)
⚫ Rate of motion (mph)
 Density (k)
⚫ Rate of traffic over distance (vpm)
 Volume (V)
⚫ Amount of traffic (vph)
 Flow (q)
⚫ Rate of traffic (vph); equivalent hourly
rate
 Basic Relationships

Low volumes Highest speeds

High volumes Lower speeds

Highest volumes Medium density

Maximum density No speed or flow

q=ku
Basic Relationships

q = ku
1
q=
h
1
k=
s

5
 Flow-Density Example

If the spacing between vehicles is 500 feet what is the density?

d = 1/k k = 1/d = 1 veh/500 feet
= 0.002 vehicles/foot = 10.6 veh/mile
 Speed-Density Relationship
Max speed 0 density

uf
Max density 0 speed
 k 

u = u f 1− 
 k 
 
Speed

Density kj
7
 Speed-Density Relationship
90
80
Average speed (mph)

70
60
50
40
30
20
10
0
0 20 40 60 80 100 120

D e nsity (p cp m p l)
8
 Flow-Density Relationship
 k 
u = u f 1 −  and q = ku
 k 
 j 
uf
 q = uf k − k2
Flow

kj

qcap Optimum density

kcap kj Density
9
 Flow-Density Relationship
3000
2500
Flow (pcphpl)

2000
1500
1000
500
0
0 20 40 60 80 100 120

10 Density (pcpmpl)
Flow – density (and speed)

qm

Slope of these
Flow (veh/hr)

lines is the space

B mean speed at
this density

0 km kj
0 KB
Concentration (veh/mi)

Do the dimensional analysis

 Flow-Density Example

If the space mean speed is 45.6 mph, what is the flow rate?
q = kus = (10.6 veh/mile)(45.6 mph) = 481.5 veh/hr
 Speed-Flow Relationship
 k 
u = u f 1 −  and q = ku
 k 
 j 
uf kj
 q = k ju − u2
Speed

uf
“Optimal” speed
for flow
ucap
maximization

qcap= kcapucap

qcap Flow
13
 Speed-Flow Relationship

90
A ve ra g e sp e e d (m p h )

80
70
60
50
40
30
20
10
0
0 500 1000 1500 2000 2500
14 F low (pcphpl)
 Highway capacity manual

Source 1985 highway capacity manual

2000 Highway Capacity
Manual
 Speed density relationship

Uncongested Flow

Queue
Discharge
Speed

Capacity Drop

Congested Flow

Flow

Source: Maze, Schrock, and VanDerHorst, “Traffic Management

Strategies for Merge Areas in Rural Interstate Work Zones
 Speed-Flow-Density Relationship
(Greenshield’s Linear Model)
uf

ucap

qcap

qcap

18 kcap kj
Special Case

Greenshield’s  k 
u = u f 1 − 

Model  k 
⚫ Linear  j 

(Only) When uf kj

ucap = kcap =
Greenshield’s 2 2
Model holds,
k ju f
qcap = kcapucap =
4
19
Greenshield Linear Model

speed (mph) uf

um

0
0 km kj
Concentration (veh/mi)
Example 1
A highway section has an average spacing
of 25ft under jam conditions and a free-
flow speed of 55mph. Assuming that the
relationship between speed and density is
linear, determine the jam density, the
maximum flow, the density at maximum
flow, and the speed at maximum flow.

21
Example 2
Traffic observations along a freeway lane
showed the flow rate of 1200vph occurred
with an average speed of 50mph. The
same study also showed that the free-flow
speed is 60mph and the speed-density
relationship follows the Greenshield’s
model. What is the capacity of this lane?

22
Example 3
A section of highway is known to have a
free-flow speed of 55mph and a capacity
of 3300vph. In a given hour, 2100 vehicles
were counted at a point along the road. If
Greenshield’s model applies, what would
be the space mean speed of these 2100
vehicles?

23
 Queuing Theory

The theoretical study of waiting lines,

expressed in mathematical terms

input server output

queue

Delay= queue time +service time

Queuing theory

 Common ways to represent (model)

queues
⚫ Deterministic queuing
⚫ Steady state queuing
⚫ Shock waves
Common Assumptions

 The queue is FCFS (FIFO).

 We look at steady state: after the
system has started up and things
have settled down.

(all the information necessary to completely

describe the system)
Typical cases where
queuing is important
 Bottle necks – capacity reductions
⚫ Lane closures for work zones on
multi-lane facilities
⚫ Toll booths
⚫ Where else would we experience a
line in traffic?
 Deterministic queuing model
 Treats the bottle neck like a funnel
 Assumptions
⚫ Constant headways through out analysis period – what
does this imply about density?
⚫ Capacity does not vary with traffic flow variable
(speed, density or flow)
Deterministic queuing treats
the waiting line as if it has no
length
 Deterministic queuing
example
50 ft 500 ft 780 ft 16 ft

F lashing Arrow Board

Data Collection T railer T aper, w/ Drums every 65 ft T ype III Barricade

Data Collection T railer

Direction of T raffic

Direction of T raffic

Example – Suppose we have a lane closure on a freeway and

the single lane capacity at the closure is 1,400 vehicle per hour.
Assume that during the first hour we expect the traffic volume
of 2,000 vehicles per hour for one hour and then the volume to
reduce to 800 vehicle per hour. At the peak, how long will the
queue be and how long will it take to dissipate?

Draw a queuing diagram (cumulative arrivals on Y axis, time on

X axis)
Deterministic queuing model
example
2,800 800 vph arrival rate C
Vehicles, Cumulative over Time
2,000

2,000 vph arrival rate

600 vehicles

1,400 vph queue discharge rate

1,000

A
1 2
Time, (hours)
Deterministic queuing
traffic signal case
Diagram for traffic signal
 Example
( R + t0 ) *  R
 = Area under the curve = delay
2
noting that ( R + t0 ) *  =  * t0 ...
R
 * t0 * = delay
2
λ µ

t0
R
Example

 = 5 vehicles per minute

R = 30 seconds or 0.5 minutes
 = 16 vehicles per minute

 = = 0.3125

0.3125 (0.5)
t0 = = 0.23
1 − 0.3125
0.23 *16 * 0.5
Total Delay = = 0.92 minutes per cycle
2
Calculating Delay

D=
((t0 *  ) R )
= total delay
2

= Average Delay
Intersection delay in both directions
Multiple directions
Minimizing delay
Steady State Queuing

 Arrival Distribution
⚫ Described by a Poisson distribution
 Service Method
⚫ Usually first come first serve
 Service Distribution
⚫ Follows a negative exponential distribution
 Number of channels
 Assumption of steady state
⚫ Average arrival rate is less than average service
rate
 The M/M/1 System

Poisson Exponential output

Process server
queue
 Steady state queuing
equation variables
• Variables

q = Average arrival rate

Q = Average service rate
n = Number of entities in
the system
w = Time waiting in the
queue
v = Time in the system
 Arrivals follow a Poisson process

Readily amenable for analysis

Reasonable for a wide variety of situations
 a(t) = # of arrivals in time interval [0,t]
 q = mean arrival rate
 →0 = small time interval

Pr(exactly 1 arrival in [t,t+]) = q

Pr(no arrivals in [t,t+]) = 1-q
Pr(more than 1 arrival in [t,t+]) = 0

Pr(a(t) = n) = e-q t (q t)n/n!

 Model for Interarrivals and Service times

• Customers arrive at times t0 < t1 < .... - Poisson

distributed
• The differences between consecutive arrivals are the
interarrival times : n = tn - t n-1
• n in Poisson process with mean arrival rate q, are
exponentially distributed,
Pr(n  t) = 1 - e-q t

 Service times are exponentially distributed, with

mean service rate Q:
Pr(Sn  s) = 1 - e-Qs
 System Features
 arrival times are independent
 service times are independent of the arrivals
 Both inter-arrival and service times are memoryless
Pr(Tn > t0+t | Tn> t0) = Pr(Tn  t)
future events depend only on the present state

→ This is a Markovian System

Example of where assumptions are
violated

4-way
Stop
Arrivals are not random

2-way
Stop
 Common one channel
equations
n
q  q
P ( n) =   1 − 
Q   Q
Example – suppose that cars take an average of 5 seconds
at a stop sign. If 9 cars per minute arrive at the sign what
is the probability of having 5 in the system what is the
probability of having five or fewer.

P(5) = (9/12)5(1-9/12) = 0.06

P(4) = 0.08
P(3) = 0.11
P(2) = 0.14
P(1) = 0.19
P(0) = 0.25
0.83
 Common queuing equations
Expected number of vehicles Expected number of Vehicles
in the system in the queue
 q   q2 
E ( n) =   E (m) = 
− 
 Q q   Q (Q − q ) 
 9   92 
E ( n) =  =3 E (m) = 

12 − 9   = 2.25
12(12 − 9)
 Common queuing equations

Average wait in the queue Average wait in the system

 q   1 
E ( w) =   E (v ) = 
 Q (Q − q )   (Q − q ) 


 9   1 
E ( w) =   = 0.25 E (v ) =   = 0.33
12(12 − 9)  (12 − 9)
 Common queuing equations
Probability of spending less than Probability of spending less than
time t in the system time t in the queue
 q
 q − 1−  qt
q
− 1−  qt
P( w  t ) = 1 − e  Q
P (v  t ) = 1 − e  Q
Q
 9
 9  9 − 1−  9*0.75
− 1−  9*0.75
= 0.81 P( w  0.75) = 1 − 12 e = 0.86
 12 
P(v  0.75) = 1 − e  12 
 Common queuing equations

Probability of having more than n vehicles in the system

N +1
q
P(n  N ) =  
Q
8+1
9
P(n  8) =   = 0.075
 12 
Shock wave theory

 Recognizes that density is a variable

⚫ Recognizes the dynamics of the traffic
flow
 Considers non-steady state condition
 More realistically represents traffic
flow
Example shock wave
Flow – density and speed

qm

Slope of these
Flow (veh/hr)

lines is the space

B mean speed at
this density

0 km kj
0 KB
Concentration (veh/mi)
Back Ward Moving Shock
wave
Qc

Speed of the
Q1
Spe
e
Shock wave
d, S

 Q1 + Q2 
S =  
Q3 Q2
Flow Rate (q)

Qb

 K1 + K 2 

k1 Density (k) k2
 Example
Suppose you have 2,000 vehicle per hour approaching a lane closure
at and average speed of 65 mph. The capacity of the lane closure is
1,400 vehicles per hour and at the maximum capacity move at 20
mph. Assuming the approaching vehicles are evenly distributed
between the two lanes. How fast is the shock wave traveling
backwards?

Q1 = 2,000 vehicles per hour

K1 = 1,000 per lane per hour/65 mph = 15.38 vehicles per mile
Q2 = 1,400 vehicles per hour
K2 = 1,400 per lane per hour/20 mph = 70 vehicles per mile
 2000 + 1400 
S =  = −11mph
 15.38 + 70 
How fast would the shock wave move backwards if all vehicle
approach in a single lane?
Forward moving shock wave

Qc

Q3 Q2
Flow Rate (q)

Qb

ed, S
Spe

Q1

K1 Density (k) K2