Chapter 5

Traffic Flow Characteristics

1. INTRO DUCTIO N
In Chapter 4, we synthesized the fundamental characteristics of the driver, the vehicle,
and the environment. In this chapter, we examine the flow of vehicles moving individ-
ually or in groups on a roadway or track, subject to constraints imposed by human be-
havior and vehicle dynamics.
It is vital to be able to design and operate transportation systems with the great-
est possible efficiency and safety. Understanding the basic principles of traffic flow the-
ory is one way of attaining this end. At the present time, there is no unified theory of
traffic flow. Much of the knowledge currently available in this field is largely empirical.
In this chapter, we treat this topic at an elementary level, with the hope that the
information will help engineers to understand problems occurring in the field, and pro-
vide the basic tools to solve these problems.

2. THE NATURE OF TRAFFIC FLOW

Traffic flow is a complex phenomenon. It requires little more than casual observation
while driving on a freeway to discover that as traffic flow increases, there is generally a
corresponding decrease in speed. Speed also decreases when vehicles tend to bunch to-
gether for one reason or another.
Thaffic flow is a stochastic process, with random variations in vehicle and driver
characteristics and their interactions. This statement needs some explanation. It is quite
common to construct models of reality in which the effects of chance variation are ig-
nored or averaged out, where any given input will produce an exactly predictable out-
put. These models are deterministic. The alternative situation is to allow random
variation in the model and then to look at the probabilities of different outcomes. This
method of stochastic modeling takes into consideration the variability among possible
outcomes, not just average outcome (Lay,l986a, 1986b).
119
 120 Traffic Flow Characteristics
Chap,1
3. APPROACHES TO UNDERSTANDING TRAFFIC FLOW
The interaction between vehicles and their drivers. and also among ve~icles, is a hi
co~pleJt process. There are three main appr?acbes to the understandtng and qua:.')
cahon of traffic flow. The first is a macroscoptc approach that looks at the flow in a if~
gregate sense. Based on such physical. analogies as. heat flo~ and fluid flow~ ~t·
macroscopic approach is most appropnate for studytn~ .steady state phenomena~
flow and hence best describes the overall operational effictency of the system.lb I
ond is a microscopic approach that considers the response of each individual vehi~l~·
a disaggregate manner. Here the individual driver- vehicle combination is exam· e,n
r
. I . h' h
SUCh as car maneuvering.This approach is Used ext~nSIVe tn lg way Safety WOrk.'fbt
'!led,
I
t~ird approach is the human-factor appro~ch. Bas1cally: It seeks to d~fine the mecba.
m~m by which an individual driver (and h1s ~r her ve~cle) locates h1mse~ or hersell
w1th reference to other vehicles and to the htghway/gmdance system. Nottce that tb
microscopic and the human-factor approaches are closely related (Drew, 1968). e
One way of combining all three appro~ches. is to a.ssume initially t.h~t a stream 01
traffic is composed of identical vehicles and 1denllcal dnvers, thus penruttmg easy inte.
gration of the various approaches. The simplest combination also assumes that the trar.
fie moves at uniform speed and that the vehicle spacing is dependent on speed. In
other words, a vehicle's behavior is forced on it by other vehicles in the traffic stream.
Indeed, speed is assumed to be the only variable that influences flow. Naturally, there is
one particular vehicle flow associated with a speed adopted by the traffic stream (Lay,
l986a, 1986b).

4 . PARAMETERS CONNECTED WITH TRAFFIC FLOW

There are at least eight basic variables or measures used in describing traffic flow,and
several other stream characteristics are derived from these. The three primary vari-
ables are speed (v), volume (q), and density (k). Three other variables used in traffic
flow analysis are headway (h), spacing (s), and occupancy (R). Also, corresponding to
measures of spacing and headway are two parameters: clearance (c) and gap (g).

1. Speed is defined as a rate of motion, as distance per unit time, generally io

miles per hour (mph) or kilometers per hour (km/hr). Because there is a broad distrib-
ution of individual speeds in a traffic stream, an average travel speed is considered.
Thus. if travel times r, t~, rl, .. . , r., are observed for n vehicles traversing a segment of
length L , the average travel speed is

(I)
where

v, = average travel ~peed or space mean speed (mph)

L = length ~f the h1ghway segment (miles)
I; = travel hme of the ith vehicle to traverse the section (hours)
n = number of travel times observed
, " '"" '"'"" ~ 1..cmnected with Traffic Flow
121
E -ple l

:nu-ee vehicles are traversing a l · mile segment or 3 highway 8 d lh r0 U . .

ts made: n e owmg observation

Vehicle A : 1.2 min-+ 0.0200 hr/mi = SO mph

Vehicle 8 : l.S min-+ 0.0250 hr/mi c 40 mph
Vehicle C: 1.7 min -+ 0.0283 hr/mi = 35.3 mph
What is the average travel speed of the three vehicles?

Solution

Average travel time .. 0.0200 + 0.0250 + 0.0283

3
• 0.0244 hr/mi
Average travel speed • 1/0.0244 = 40.91 mplt.

The average travel speed calculated is referred to as the space mean speed. Jt is called
"space" mean speed because the use of average traveltime essentially weights the average
according to the length o f time each vehicle spends in "space."
Another way or defining the "average speed" or a traffic stream is by finding the
time mean speed (v,) . This is the arithmetic mean or the measured speeds of all vehicles
passing, say, a fixed roadside point during a given interval of time, in which case, the indi·
vidual speeds are known as "spot" speeds.

(2)

where V; is the spot speed, and n is the number or vehicles observed.

Example2
Three vehicles pass a mile post at 50, 40, and 35.3 mph, respectively. What is the time mean
speed of the three vehicles?

Solution

•
LVI _ so + 40 + 353·· 41 n h
,.,
Average time mean speed == v, = - - - "' · mp
3
11

It can be shown that whereas the time mean speed is the arithmetic mean o~ the spot
s ed the s ace mean speed is their hannonic mean. Time mean. speed ts always
;ate~ than ~pace mean speed except in the situation where all vehicles travel at the
same speed. It can be shown that an approximate relationship between the two mean
speeds is
 Traffic Flow Characteristics
122

Also,
q~
v,= v, - v,
. ~~
where u 2 is the variance of the space mean speeds. For example, tf 11, is known
41.77 m~h and the variance, u: = (I (111 - 11,j1~/n, is equal to 37:58, then frorn . to lit
~lla.
tion (4) , 11s can be found to be 40.91 mph, and thts result checks wuh the figure .tnd,ca
ed in Example 1. t.

2. Volume and rate of flow are two differ~nt meas.ures. V,olume.is the actual nu
ber of vehicles observed or predicted to be pas~mg a POt_Dt durmg a g..v~n time inter,rn.
The rate of flow represents the number of vehtcles passmg a pomt dunng a time . ~
. h Inter.
val less than 1 hour, but expressed as an eq01va1ent our1Y rate. Thus, a volu
200 vehicles observed in a 10-mmute . pen'od .IIDP lies a rate o f flow of (100lilt 01
60)/10 = 1200 veh/hr. Note that 1200 vehicles do not pass the point of obsel\fa/
101
during the study hour. but they do pass the point at that rate for 10 minutes.
Example3
Calculate the rate of flow of vehicles from the following data:

llmcperiod Volume (vehicles)

4:00-4:15 700
4:16-4:30 812
4:31-S:OO 1635
Total 3147

Solution
~hhouf:h the volume = 3147 vehlhr, the individual rates of flow during illt
rhree t1me penods are 2800,3248, and 3270 vehlhr, respectively.

. 3. Density or concentration is defined as the number of vehicles occupying 1

gt~en length. of lane or roadway, averaged over time, usually expressed as vehicles pel
mtle (vehlmt). Dtrect measurement of density can be obtained through aerial photog·
raphy, but more commonly it is calculated from Equation (5) if speed and rate of floll'
are known.

q = vxk (~
where
q = rate of flow (vehfhr)
11 = average travel speed (mph)
k = average density ( veh/mi)
•·
;rc( . 4 Parameters Connected with Traffic Flow 123

Thus. a highway segment with a rate of flow of 1350 vehlhr and an average travel
speed of 45 mph would have a density of k = (1350 veh/hr)/45 mph = 30 veh/mi. The
proximity of vehicles in a traffic stream is given by density, which is a critical parameter
in describing freedom of maneuverability.

4. Spacing and headway are two additional characteristics of traffic streams.

Spacing (s) is defined as the distance between successive vehicles in a traffic stream as
measured from front bumper to front bumper. Headway is the corresponding time be-
tween successive vehicles as they pass a point on a roadway. Both spacing and headway
are related to speed, flow rate, and density.

A vg. d enstty 80. :.:.,~ft~/m:.:.i-:-:--:-

. (k) , vehlmt. = __ _:5:.:2.:::; (6)
avg. spacmg (s), ftlveh
avg. spacing (s), ftlveh
Avg. headway (h), sec/veh = ( ) ftl (7)
avg. speed v , sec
3600, seclhr
A vg. flow rate ( q ), vehlhr = d h(h) cl h (8~
avg. ea way , se ve
Spacings of vehicles in a traffic lane can be generally observed from aerial pho-
tographs. H eadways of vehicles can be measured using stopwatch observations as vehi-
cles pass a point on a lane.
5. Lane occupancy is a measure used in freeway surveillance. Hone could mea-
sure the lengths of vehicles on a given roadway section and compute the ratio

R = sum of lengths of vehicles = 'i'. L1 ( )

9
length of roadway section D
then R could be divided by the average length of a vehicle to give an estimate of the
density (k).

Eumple4
Four vehicles, 18, 20, 21. and 22 ftlong. are distributed over a length of a freeway lane 500
ftlong. What is the lane occupancy and density?

Solution

- 18 + 20 + 21 + 22 = 0.162
R - 500
Average length of vehicle = 20.25 ft
5280 hi .
k = 0.162 X . = 42.24 ve . m•
20 25

lAne occupancy (LO) can also be described as the ratio of the time th~t vehicles are
present at a detection station in a traffic lane compared to the ttme of sampling.
total time vehicle detector is occupied It0 {10)
LO = total observation time =T
 Traffic Flow Charact .
124 l!nst;~

Direction of ~ravel

f-c-J
1.. • Length or vehicle
c• Dinance between loops or detector

Figure 5-1 illustrates the use of a detector in traffic engineering work. Here

L +C
to =

where Lis the average length of vehicle, and Cis the distance between the loop or tbedt.
lector.
It is necessary to know the effective length of a vehicle as measured by thed
in use to calculate lane occupancy. Density can be calculated by using the express:altaq

LOX 5280
k = L+C (II)

In most cases. the detector is actuated as soon as the (ront bumper crosses tbedetector&llll
remains on until the rear bumper leaves tho detector.

Example S
During a 60·sec period. a detector is occupied by vehicles for the following times:O.ll,
0.38, 0.40, 0.32, and 0.52 sec. Estimate the values of q, k, and v. (Assume that the loop-<!<·
lector length is 10 fl and that the average length o f vehicles is 20ft.)

Solution

:Eto = 0.34 + 0.38 + 0.40 + 0.32 + 0.52 = 1.96 sec

n=S
100%
Lo = 1.96 X "'6{) • 3.27%

The average effective length of a vehicle plus distance between loops is assumed to l<
20 + 10 • 30ft; then

3.27(5280)
k = lOO 30 = 5.75 vehlmi
sec.4 Parameters Connected with Traffic Flow 125

n(L +C) S x 30
v, ; :Ero s 1.% • 76.53 ft/sec = 52.18 mph

q= k X v, = 5.75 X 52.18 c 300 vehlhr

It must be noted that in this example use is being made of an average effective length
of a vehicle plus distance between detector loops totaling 30 ft. This assumption can
lead to serious error, because the mixture of cars and trucks in a traffic stream can ap-
preciably change the effective length of vehicles used in the calculations (Gerlough
and Huber,1975).
Lane occupancy applications are of much concern in on-line surveiUance and
control systems. For example, freeway surveillance is accomplished by monitoring lane
occupancy from numerous stations such as entrance ramps. Proper records of lane oc-
cupancy can be useful in evaluating traffic stream perfonnance.
6. Clearance and gap correspond to parameters of spacing (ft) and headway
(sec). These four measurements are shown in Figure 5-2. The difference between spac-
ing and clearance is obviously the average length of a vehicle in feet . Similarly, the dif-
ference between headway and gap is the time equivalence of the average length of a
vehicle (Uv ):

g=h - (~) (12)

and
c = gxv (13)

where
g = mean gap (sec)
L = mean length of vehicles (ft)
c = mean clearance ( ft)
h = mean headway (sec)
v = mean speed (ftlsec)

Example6
Figure 5-E6(a) shows a time-space plot of vehicles within a time-space domain. Based on
the information contained in this figure, several parameters can be calculated.
L An observer counts six vehicles c.rossing line A-A' in 25 sec. Calculate the flow of ve-
hicles per hour (q).

1--
I Clearance ( f t ) - + - L (It)
Gap(sec}
-i
l I
~r-- l:
Spacing (ft) or headway (sec)---\
~ Flpre 5-2 Clearance-Gap and
Spacing· Headway Concept.
 Traffic Flow Characttristic:s
126
~D. I

~
A•
.:::
~

..8
c
:;
iS
(10)

(U)
(12)

0
Time (sec)
(a)

18.2

12.1
(b)

Fie- S.E6 Data for Enmple 6: (a) Tune.SJ*e Plot ofVebicle Paths; (b) Headways.
Sec. 4 Parameters Connected with Traffic Flow
127
n 6veb
q• T ~ -25 ~ 0.24 veblsec • 864 veblbr
sec

b. A timer is ~gun at time 0.0 sec continuing for 25 sec and the times at which the front
end of veh•cles pass the observation point A- A • are noted as folio s [ Fi 5
E6(b)]: w see •gure -

Time of Pa.uing
Vehicle (su)
4 3.0
5 5.2
6 8.2
7 12.1
8 lS.S
9 18.2

What are the individual headways and tbe average headway? Because there are six ve-
hicles, only tbe first five headways can be determined directly.

h,_ 5 = 2.2
hs-6 = 3.0

~-7 = 3.9

h 1-l = 3.4
ha -9 = 2.7

~-· = 9.8
Total= 25.0

The final headway can be calculated as shown in Figure 5-E6(b):

Average headway, h = 625 • 4.17 sec

or h = 36001864 = 4.17 sec

c:.. A speed trap consisting of a record of times when vehicles passed two poi.nts, AA' and
88', was made as indicated on figure 5-E6(a). find tbe average speed of vehicles 4, 5,
6, 7, 8, and 9, assuming that the trap distance is 240ft.

16 03
Time-mean speed v, = ~ • 26.84 mph

6 X 240
Space-mean speed v, = . • 37.11 ftlsec = 25.30 mph
38 8
128 Traffic Flow Charact~rlstics
~·\
Vehicle T~~ne of Passing TrapTtme
Ttme of Passing (sec) Speed
AA' 88' (rnp~)
4 3.0 II.S s.s 19~
s S.2 13.1 7.9 20.7;
6 8.2 1S.2 7.0 23.38
7 12.1 18.1 6.0 27.27
8 IS.S 20.1 S.2 31.46
9 18.2 22.4 4.2 38.96
Total 38.8 161.03

Note that in calculating the space-mean speed, the numerator is the tota:;----
traveled by the six vehicles within !he roadway le~gth and the denominator is t~t~~~ct
time these six vehicles spent on thts roadway sectton. I~
d. At the time of observation. there were seven vehicles coumed on the IOOO.rt lec:tiatt
roadway lane. Calculate the density (k). ~

7 veb X 5280 ft/mi hi .

k• • 36.96 ~ mt
1000 ft
e. 1Wo aerial photographs were taken, 2 sec apart. and Ihe pc_>Sitions of vehicles 1Uuoo.p
7 as shown in Figure 5-E6(a) were no1ed.These seven vehtcles are tbesame vebidca
those observed in section (d) of this example. Calculate Ihe average speed andn~
flow.

Vehicle Position I Position 2 Feet

I 680 160 8Q
2 610 670 60
3 S60 615 ss
4 4SO SIO (j()
s 38Q 440 (j()
6 210 3SO 80
7 60 160 100
Total 495
495
v,- m- 35.35 ft/sec = 24.11 mph
q = kv • 36.96 x 24.11 = 891 veh/hr

f. Consid.er ,'he tim.e-space diagram (Figure 5·E6(a)J and Ihe lrajectories of the 12 .tbi-
cles. Wttbm the tuue"space domain, calculate Ihe flow, densily, and speed.

q = Ix, lc • It1 :Ex1

A A v, = :Et,
where

x, = distance traveled by the ith vehicle within the time-space domain

1
1 = time Iaken by the ith vehicle to traverse the time-space domain
A • area of the time-space domain
A • IOOO X 2S • 25,000 ft-sec
7000veh-fl
q a 25,000 ft-sec = 0.28 vehlsec = 1008 veh/hr
sec. 5 Categories of Traffic Flow
129
194 veh-sec
k = 25,000 ft-sec = 7.76 x to-l veh/ft = 40.97 veh/mi

7000 veh-ft
11
• = 194 veh-sec = 36.08 ftlsec = 24.60 mph

Vehicle, i Distance Ttaveled1 x; lime Taken, 1

(veh-ft) (sec)
I 400 10
2 440 15
3 480 20
4 600 21
5 680 22
6 800 23
7 1000 25
8 920 20
9 880 15
10 360 10
11 240 8
12 200 5
:1;x1 =7000 It = 194

5. CATEGORIES OF TRAFFIC FLOW

Vehicle flow on transportation facilities may be generally classified into two categories.
1. .Uninterrupted flow can occur on facilities that have no fixed elements, such as
traffic s1gnals, external to the traffic stream, that cause interruptions to traffic flow.
Traffic flow conditions are thus the result of interactions among vehicles in the traffic
stream and between vehicles and the geometric characteristics of the guideway/road-
way system. Also, the driver of the vehicle does not expect to be required to stop by
factors external to the traffic stream.
2. Interrupted flow occurs on transportation facilities that have fixed elements
causing periodic interruptions to traffic flow. Such elements include traffic signals, stop
signs, and other types of controls. These devices cause traffic to stop (or significantly
slow down) periodically irrespective of how much traffic exists. Naturally, in this case,
the driver expects to be required to stop as and when required by fixed elements that
are part of the facility.
It should be noted that uninterrupted and interrupted flow are terms that de-
scribe the facility and not the quality of flow. A congested freeway where traffic is al-
most coming to a halt is still classified as an uninterrupted flow facility, because the
reason for congestion is internal to the traffic stream. A well-timed signaling system on
an arterial may result in almost uninterrupted traffic flow, but such a flow is likely to be
interrupted for several reasons, primarily as part of the system, and is therefore classi-
fied as interrupted flow (TRB, 2000).
Table 5-1 provides a sample of the types of facilities under the categories of unin-
terrupted and interrupted flow facilities. Note that the categorization is approximate.
Freeways, for instance, operate under the purest form of uninterrupted flow, and multi-
lane and two-lane highways may also operate with almost uninterrupted flow, particu-
larly in long segments between points of fixed interruption~, such as segments where
 130 Traffic Flow Characteristiu

TABLE 5·1 Types of Transportation Fadlitles

Uninterrupted Dow
Freeways
Mullilane highways
1\olo-Iane hi&hways
Interrupted Row
Signalized streets .
Unsignalized streets with stop s1gns
Arterials
'Danslts
Pedestrian wolkways
Bicycle paths

signal spacing exceeds 2 miles.. Pedestrian, bicycle, and transit flow are gencraU
sidered to be interrupted, although uninterrupted flow conditions can OCcur-~rtlt.
ample, in a long busway, without stops. tt.

6. THE UNINTERRUPTED TRAFFIC FLOW MODEL

This model can best be described by means of a typical curve of the form showa il
Figure 5-3. Imagine several vehicles, driven by rational drive~ along a section offre~.
way. As vehicles speed and spacing increases, the speeds approach the Cree speed, and
drivers adopt their own speed when uninfluenced by other vehicles in the trtf!k
stream (point C). It is useful to unders tand the situation at point A , which repre$e0b
the maximum traffic density that occu~ when traffic has virtually come to a complck
st.op, reminiscent of a linear parking loL The dashed curve represents the normal !lor
behavior if all drivers we re to have the same free speed (point D). It bas been obstrved
that drivers are uninfluenced by other vehicles in the traffic lane at flows of about900

Normal now

:!-----+--
1
Forced now
I
I

Congestion Capacity
Vehlbr
sec. 1 Analysis of Speed, Flow and Density Relationship
131
veh/hrllane or less, which is about half th ·
forced-flow region, each vehicle ado t . e ~a~acJty flo~ (point B). Note that in the
P s 1ts nurumum spacmg and clearance distance.

1. ANALYSIS OF SPEED, FLOW, AND DENSITY RELATIONSHIP

If it is hypothesized that a linear relat" h" ·
uninterrupted traffic lane and the tr~~~sdio~~~s~s b~w~)en th~ spee~ o~traffic on an
then mathematically this relationship can be repres::te':b/s s own m •gure 5-4(a),

11 = A- 8k
where
mean speed of vehicles (mph )
11 =
k = aver~~e density of vehicles (veh/mi)
A, 8 = empmcally determined parameters
Because the flow of a n uninterrupted traffic stream is the product of the density
and the speed, we have
q = kv = Ak - 8k2 (14)
(v - A )v A if
q = kv = =-:::-v- -
- 8 8 8
At almost zero density, the mean free speed equals A , and at almost zero speed,
tbe jam density equals A/8. The maximum flow occurs at about half the mean free
speed and is equal to K-148.
Figure 5-5 shows the theoretical relationship between flow (q) and density (k) on
a highway lane, represented by a parabola. As the flow increases, so does the density,
until the capacity of the highway lane is reached. The point of maximum flow (q.,..)
corresponds to "optimal" density (k0 ). From this point onward to the right, the flow
decreases as the density increases. At jam density ( ki), the flow is almost zero. On a
freeway lane, this point may be likened to the traffic coming to a halt, where the lane
appears to look like a partcing lot.
If rays are drawn from zero through any point on the curve, the slope of the rays
represents the corresponding space mean speed. The ray with a slope of vr corresponds

~
~
~
.., -a
-
~
Q.
s ....," !.
l r-,.-{2-
~

• • A
.
,;
§.
A
~
>
...
i
..lJ
~
Ci: ~
Density, Jc (""b/mi) Flow, q (veh/hr)
Oensity.Jc (veh /mi)
(b) (c)
(a)

Flpre 5-4 Speed-Aow-Oensity Curves.

 132 Traffic Flow Charact- . .
dtSti(S

.,

0 ko
Denshy. k (•eh tmi) F"IJIIre 5·5 Aow·Densily Cur-e.

to the mean free speed and is tangential to the curve. This speed is possible when~
density is near zero.
Figure 5-6(a) shows a theoretical relationship between speed and density repn.
sented by a straight line. This relationship is not really true but is quite useful in practkt
Aows can be calculated simply by multiplying coordinates or speeds and densitiab
any point 011 the straight line.
Finally, Figure 5·6(b) shows the theoretical relationship between speed and Dot.
Rays drawn from zero to any point on the curve have slopes whose inverse is equaho
the density.
The hypothetical diagrams connecting mean speed, density, and now shown iD
Figure 5-7 and the corresponding conditions on the road can best be discussed as fol-
lows. At point A, density is close to zero, and there are only a very rew vehicles on tbe
road; the volume is also close to zero and these £ew vehicles on the road can chOOR

Are••,Pk, O
represents lhe
!low for speed
""and dena:ity k~

0 k, k; k 0
(1) (b)

flpre s-6 Speed-Density and Speed·Aow CurveL

 sec. 8 Empirical Studies of Traffic Stream Characteristics
133

•
l"
~ c
D
Density,k D Flow.q

c
...
~
£
A D flcurt 5-7 Curves Showing the
Density, k Connectioos Betweca Mean
Speed, Density. and Flow.

their own individual speeds, or change lanes with no restrictions. At point B, the num-
ber of vehicles has increased but the conditions are of "free flow'' and there are hardly
any restrictions, although such restrictions keep increasing steadily by the time point C
is reached. From B to C, the Oow conditions may be catted "normal," but as density in-
creases, drivers experience significant lack or freedom to maneuver their vehicles to
the speed and lane of their choice. Around point C, traffic conditions begin to show
signs of instability, and speeds and densities nuctuate witb small changes in volume.
Point C is the point of maximum volume, and further increases in density reduce
speeds considerably. Such behavior is called forad flow and prevails all along from C
almost to point D. Flow near point D is reduced almost to zero, witb cars stacking up
almost bumper to bumper. Point D is known as jam density. A driver would perceive
excellent driving conditions from A to 8 , moderately good conditions from 8 to C, but
increasingly deteriorating conditions from C to D.

8. EMPIRICAL STUDIES OF TRAFFIC STREAM CHARACTERISTICS

8.1 Macroscopic Models of Traffic Flow

(a) Greenshlelds' Model The general model connec!ing speed, flow, and densi-
ty discussed 50 far is a linear model proposed by Greeosbields (1935). As field mea-
surements of speed, flow, and density became available, several researchers evolved
 Traffic Flow Characterlsti<s

traffic flow models based on actual curve fitting and staustl

. 'cal .
testmg. lbe ev
~-\
of models proceeded along two lines: al~qtia~
1. Relationships of q-k·v were tested in_ terms of.goodness of fit to a:tual litld d
2. Relationships were supposed to sat1sfy certatn boundary cond111ons (se lta.
Figure 5-4): _ · e a~
•- Flow is zero at zero density.
b. Flow i.s zero at maximum density.
(. Mean free speed occurs at zero density.
d. Flow-density curves are convex (i.e., there is a point of maximum flow).

v, m v1 - (~)k (IS)
Greenshields' model provides the slope and intercept by hand-fitting a 5 •
line to plotted data or by using linear regression.This model satisfies alit~
boundary conditions, although the statistical quality may be poor (e.g., low 1
ficients of determination and high standard errors). I:Qef.

(b) Greenberg's Model Greenberg (1959) developed a model taking 5

flow, and density measurements in the Lincoln Thnnel resulting in a speed~
model. He used a fluid-flow analogy concept, using the following form:

v, = Cln (k/k) where C is a constant

(16)
Substituting qlk for v,

q = Ckln(k/ k) (17)
= Ck(lnk1 - Ink)
Differentiating q with respect to k, we obtain

dqldk = C[k(-1/k) + (lnk1 - tnk)] = 0

And for maximum q,

C(- l + lnk1 - Ink) = 0

Therefore,

lnk1 - Ink =t (IS)

and

ln(kf k ) = 1 (19)

Substituting Equation (19) in Equation (16),

11, (maximum) =C (20)

soC is the speed at maximum now.
 empirical Studies of Traffic Stream Characteristics 135
Se<- 8
Greenberg's model shows better goodness-of-fit as compared to Greensbield's
model, although ic violates the boundary conditions in that zero density can only be ac-
tained at an infinitely bigb speed.

E,.....ple7
The speed-density relacionship of craffic on a section of a freeway lane was estimated co be

v, ~ 18.2ln(220/k)

•· What is lhe maximum flow, speed, and density at chis flow?

b. What is lhe jam density?

Solution
a. q = vk = 18.2kln (220/k) ~ 18.2k(ln220- Ink)

Maximum Dow occurs at dq/dk ~ 0. Therefore,

lnki - Ink = 1
ln220 - Ink ~ 1
5.394- Ink = I
k = 80.93 veh/mile, when q is maximum
q.,.., = (18.2} (80.93 ) (I) ~ 1473 vehlhr and v = 1473/80.93 = 18.2 mph
b. Jam density is obviously 220 veh/mi.

Eumple8
Given k = 130 veb/mi; and k = 30 veh/mi, when v, = 30 mph. Find q...,.
1

Solution

v, = Cln(k/ k }

At k = 30 veh/mi, v, ~ 30 mph. Therefore,

30 = Cln ( !30130)
C =30/1.466 = 20.459 =v, at ma.ximum flow
v, = 20.459ln (kjk)

For maximum flow (q),

tnki- Ink =1
tnl30- Ink= I
136 Traffic Flow CharactA . .
-nstoes
or

Ink= 3.8675
k = 47.82 veh/mi
q.... = (20.459)(47.82)(1)
= 978veh/hr

8.2 General and linear Speed-Density Relationship

Figure 5-8 shows a series of plots connecting flow and speed, spacing and spe d
ty and flow. Note that the basic assumption is that the density-speed relations~· ,defl!i.
linear. Contours of equal spacing, equal flow, and equal speed are superimpoJ 1111q
appropriate figures based on the fundamental equation, spacing = speed/Dowon lht
the appropriate points of jam density and maximum (or optimum) flow are ind-' Also,
· · between denstty
If a I.mear reIahonship teat!()
· an d speed ·IS assumed, the correspo ,r,,~
results are shown in Figure 5-9. As mentioned before, this assumption was ori;l~
made by Greenshields (1935) on the basis of empirical data. Note again that inn~y
case, the flow is a parabolic function of speed and that the maximum flow occursatb~
the free speed.

Example9
Assuming a linear speed-density relationship, the mean free speed is observed to be 81
mph near zero density, and the corresponding jam density is 140 vehlmi. Ass.ume tlt.ltlhe
average length o[ vehicles is 20 ft.
a. Write down the speed-density and flow-density equations.
b. Draw the v-k, v-q, and q-k diagrams indicating critical values (Figure 5-E8).
c. Compute speed and density corresponding to a flow of 1000 vehlhr.
d. Compute the average headways, spacings. clearances, and gaps when lhe now a
maximum.

Solution

a. The equation for the linear relationship of v and k is

v=mk+C
-owlltn
where m and Care constants. Substituting k = 140 when v = 0, and k -
v = 60, we get v = 60 - 0.43k:

q = vk = ( 60 - 0.43k )k = 60k - 0.43k2

·w~·
b. With q = 60k - 0.43k 2, to find the maximum value of q, differentiate q fl1
spect to k and equate to zero.

dq
dk=60 - 0.86k=O
 138 Traffic Flow Characte rist ics
Cllap
.s
Free X jam density
4

FLOW

Op Free
Speed
(a)

Asymptotic
SPACING

traffic jam
Free Free
2
Speed
Jam (b)
density

density
DENSITY at capacity
,; fij;;d.-;.;i;;.):
I
I

Free Free
2
Speed
(c)
am density

DENSITY
ig~~~·~~II~-- --

capacily
Plow
(d)
Free X jam density
4

SPACING
~~~~!f~cl~m)____ _
traffic jam

capacity
Flow Figure 5·9 Speed·Fiow.Spacing· DensiiY
(e) Relationship Based on Free Speed and a
Linear Speed-Densily Curve (Lay, 1986a).
sec. 8 Empirical Studies of Traffic Stream Characteristics
139

19 120 140 2100

Density, k (veh/mi) Flow, q (vehll\r)

-;:;- 2100

i...
~

J 1000
~

19 120
F'apro S-E8 q-lc-v Curvet
Density, k (veh/mi) for Example 8.

Therefore,

k • 70veh/mi

and
q.,. = (60 x 70) - (0.43 x (70)') = 2100 vehlhr (see figure 5-E8)

~ If q = 1000 vehlhr, 1000 = 60k - 0.43/c'. Therefore, k = 120 vehlmi or 19 vehlmi.

Also,v = 60 - 0.43k, so substituting values of k results in

v = 8.4 mph or 51.8 mph

3600 3600
d. q - -h-; therefore, h =
2100
• 1.714 sec

s 5280
h = ;: therefore,s a 1.714 X 30 X 3600 = 75.42 ft
L 20 X 3600
30 x 5280 1.26 sec
g = h - -;; - 1.714 - •

5280
c = gv = 1.26 x 30 x 3600 • SS.44 ft

Refer to figure 5-2 for def.nition of clearance, gap, spacing. and beadway.
See. 8 Empirical Studies of Traffic Stream Characteristics 141

Therefore,

T = T. - n., (25)
"' q
where q is given by

q= n,. + "•
T., + T.
Ex-pie 10
A student riding his bicycle from campus on a one-way street takes 50 min to get home, of
which 10 min was taken talking to the driver of a stalled vehicle. He counted 42 vehicles
while he rode his bicycle, and 35 vehicles while he stopped. What are the travel time and
flow of the vehicle stream?

Solution
Assuming that the vehicles counted while stopped represent the vehicle now:

q= ~~ • 3.5 veh/min = 210 vehlhr

n,. . 42 veh
T = T. - - = 50 mm - = 50 - 12
"' q 3.5 veh/min
= 38min

Example ll
The data shown were noted in a traveltime study on a 2-mile stretch of a highway using
the moving-vehicle technique. Determine the traveltime and now in each direction on this
section of the highway.
Number of Vehicles

Direction Travel lime Traveling in Ovenaking Ovenaken

(min) Opposite Direction Test Vehicle by Test
Vehicle

Nonhbound (NB) T,., N,.,

I 3.20 75 ""3 1
2 2.80 80 2 2
3 3.25 85 0 1
4 3.01 70 2 I

Average (NB) 3.07 n.s 1.75 1.25

Southbound (SB) r, Ns
"•4
I 3.30 78 0
2 3.25 74 2 2
3 3.40 79 0 2
4 3.35 82 3 3
Average (SB) 3.33 78.25 2.25 L75
 TraffiC Flow Characterlstiu
tl,
'I>.I
142

Sol<tllon
(N + ~,.y(Ts + T,)
AowNB • q1<1 • s _ t.2SJY(3.33 + 3.07) = t2.30vebJmin
• [78.25 + ( t.?S
(1'1. + ~,y(T,. + Ts)
AowSB • q., • " _ ?SJY(J.07 + 3.33) • 12.19 vebJmin
- [71.5 + (2.25 I.
. rtbbOund, i", • Til - n,lqNB
A ven~ge travelurne. no 029 •
• 3.07 - ( 1.75 - 1.25)112.30 = 3. nun
·,rne southbound, i"s • Ts - nslq,,.
Average trav.lt • .
- 3.33 - (~~
~• I 75"'? 19 • 3.289 nun
- • ,.~

9. TRAJEGORY DIAGRAMS
. . hi examined in previous sections of this chapter
The speed·fl~w-deos•ty relatJon:eh~ior of vehicles in a traffic stream. If; however~
coos1dered weth the_aggregate . f . di 'dual vehicles, a trajectory diagr '
would like to examme the behaveor o Ill VI • • • • am ca.
provide useful informaeion. An example of a traJectory deagram ·~ shown on Fi&urt
S-IO. Each ttajeclory represents the movement tbro~gb spa~e a nd time of a panicular
vehicle, and the combination of lines i!Justrat~ the ontera~toon between v~hocles. '!lot
slopes of the trajectories are the speeds of v~bocle_s. Such doagrams ~ave. wede &I'Pfic1.
tion in the srudy of platoon formation and dospersoon, a nd the coordonat10n or tralf~t.

9.1 Shock Waves and Bottlenecks

Consider a two-lane rural highway with a speed limit of 45 mph in rolling terrain where
vehicles are prohibited from passing. Under this condition, it is not uncommon for 1
slow-moving truck to be seen crawling uphill atl5 mph, with a platoon of vehicles fol-
lowing the truck. Any additional vehicles joining this slow-movi ng platoon will be pos-
sibly having an approach speed of 45 mph, but may suddenly have to reduce their

Time
Figvre 5-10 1ime-Space Oiag11l!'·
Sec. 9 Trajectory Diagrams 143

speeds down to 15 mph, creating a "shock wave"-a sudden compression of vehicles

that can lead to a rear-end collision if a driver is inatlentive.
Now, suppose that the truck in question, with a platoon of impatient drivers be·
hind it, negotiates the crest of the curve and is traveling downgrade with a speed of 45
mph. Here there is the possibility of another shock wave, this time of decompression.
A similar shock wave is likely to occur when, for example, a three-lane freeway
is reduced to a two-lane freeway and a bottlenec.k situation arises. A reverse shock
wave may occur after vehicles have negotiated the bottleneck, and the released con-
ditions aUow the compressed platoon to disperse, gain speed, and increase their
headways.
Vehicular streams interrupted by traffic signals also represent a situation where
vehicles approaching a stationary platoon are suddenly compressed (almost to jammed
density), and then released when the signal turns green, representing decompression.
A generalized bottleneck formation is shown in Figure 5-11, in which the normal
and bottleneck flow-density relationships are superimposed. A vehicle on the main
road (flow qA) with speed vA (represented by the vector OA) approaches the bottle-
neck section of the main road. Vehicles in this section have a speed of v8 (flow q 8 ), rep-
resented by vector OB. The propagation of the shock wave (chord AB) in the platoon
of vehicles foUowing the lead car is represented by vector AB, with the result that ve-
hicles approaching the bottleneck are forced to reduce their speeds to 0 B, which is
called the crawl speed. This speed moves upstream.
When changes in flow are occurring, the changes through the stream of vehicles
travel at a velocity, called the wave velocity, given by
dq (26)
u,. = dk

and if
q 8 = platoon flow (vehlhr)
k 8 = platoon de nsity ( veblmi)
q,. = free flow

Maximum
now
Original Crawl
now speed

Bottleneck
now Shock
speed
FICUft 5· U Shock Wave
Measurements.
 Traffic FloW Characteristia
~i
144

and k~ ~ free densitY

then qJk8 ~VB

and
qik· == v.

Therefore. Qa - qA
,._ ~ ks - kA (2~

where u,. is the speed of the shock. wav;:tive, the wave is proceeding in the directiOn or
If the sign of the shock wa.~e: P? ~ is negative. the shock wave is moving a&ail!t
the stream Row (downstream); 1 t. e srg shock wave exists if u ., 0. =
the stream Oow (upstream). A stattonary

Example 12
l mile kngth is located on a 40-mph highway. Stream mc•1un
A school'zone (20 mph) or •· · d· d ·
• ••01•0115 upstream middle of the school zone, an JUSI ownstnamot•'·
ments at vartous- 1 • ._
school zone, n:spectively. are u follows:
Approocltingzone: q• • JOOOveh/hr, "• • 40 mph
Middleohone: q8 = 1100 veh/hr,v8 • 20 mph
Downstream of zone: qc • 1200veh/h.r, Vc ~ 30 mph
Sketch the q·k·v curves and indicate critical values. Calculate the intensity and dirtaia&
of the shock w~ves c.reated by this speed zone. What is the length of the platoon eteatedbJ
the speed zone and the time required to disperse it? (Assume that the speed-woe reuit>
don oper>tes for only 15 min during morning and evening periods.}

Solution
See Figure S·E12.

q. • 1000 veblhr v. 40mph

q
k, • 25 veh/mi
q, • 1100 vcblhr v, = 20mph ks " 55 veblmi
qc • 1200 veblhr
Vc a 30 mph kc • 40 veh/mi
It is obviousthallhere iJ a plat . .
ol20mph. Jn Olber"'"Ords.the! olo;;:' workrng ns way through the school zone at a~
highway. By the same token, th~ d ~platoon moves at + 20 mph forward relative tothl
e 0 the shock wave at the rear of the platoon is

"•• ~ 1100 - 1000 100

55 - 25 ~ + 30 ~ 333 mph
(moving downstream or in the d"
lfection o( the stream flow).
Sec. 9 Trajectory Diagrams
145

60
so

20
10

o L-----,~~----~so~----~7~5----~,oo~
lc (vpm)

Shock wave
(front or platoon)

1200
Shock wave
1100 (rear of platoon)
1000

soo

0 ~ so 75 100
k (vehlmi) Flpre 5·E12 q·k· v Curves for Example 12.
 Traffic Flow Characteristl
146 C1 Ch
. • iCI.s
relative growth of the platoon IS proportional to the re .
Therefore. the_ mph. The platoon grows at the rate of 16.67 rn blah~e ~
16 67
between 20 - 3.33 ~ · e operates for only 15 minutes every morning. th: le~ 11'
forward. If the schoc;j ~~ 17 miles. and the number of vehicles in this platoo0 ,_"&Ill Of t•
platoon = 16.67 X oo · "4.17 ~ ·•
-m
- • k
A shoe wave
also develops where the school zone ends and the relea- (
. . - Or d.~
'
. ) ...g • s.The speed of this shock wave 1s """'~~~·
press1on zone "" 111
1200 - llOO = -6.67 m h
usc 40 - 55 p

(moving upstream or opposite to the direction of the stream flow).

Note that tbe shock wave at the rear of the platoon moves downstream at +333
Naturally the relative speed of the two shock waves is -6.67 - ( +3.33) = 10m"'- ,llll't
therefore•take 4.17nO .. 0.417 hour, or 25 mill
. Ute$. to d'ISSipate
. totauy the platoonfClrlllt!d,
""'""'

EX11nple13
Draw a time-distance sketch showing the front and rear of the platoon at any time rr0111
tbe start of 1hespeed-zone enforcement until such time as the platoon has completely"-
persed. based on Example 12.

Sol•tioo
Figure 5-El3 plots the front and rear of the platoon. The algebraic distance belloeto
the two reprcsentli the length of the platoon. Notice that the maximum lengdl h
5.00- 0.83 = 4.17 mi. Also note that the platoon is finally dispersed 40 min (IS + lS)
after the IS-min speed e nforcement, and 2.22 mi from the beginning of the l·mile·lao'
school zone.
9.2 Shock Wave Propagation
Assuming Greenshields' traffic Oow model, v1 represents the mean free speed and k,
the jam density.

v, =v 1(1- :;)
aod if

then

v, = v1( I - x)
U there are two regions in a traffic stream flow having k/ki values of x 1 and Xz. tbeB
V1 = Vt(1 - x1) and ~ = v (I - x2)
1
and
Sec.. 9 T(ajectory Diagrams 147

20mpb

]: 3
g
c
.0;;•
2.22 (40 mio.222 mi)
2

M7mpb

30 40 so
1ime (mtn)
3.3Jmpb

f"tpre S·£13 1lme-Oistanc_

e Sketch (or Example 13.

where ,.., is the ve locity of the shock wave in terms of the two densities in the corre-
sponding two regions. This test equation is useful in examining the shock wave in three
common settings: small discontinuities in density, stopping situations, and slllrting
(Homburger,1982).

9.2.1 Shock Wave Caused by Nearly Equal Densities If x1 and x2 are nearly
equal, then
U w o v1(1 - 2x1 )

This shock wave is referred to as a wave of discontinuity.

9.2.2 Shock Wave Caused by Stopping Here the density upstream x1 is brought
to a jam density condition and x2 o I. Therefore,
u., =v1{1 - (x1 + 1)] =- (v1) (x1)