CENG-431: Traffic Engineering

Chapter 11.2
Lecture 6: Spot Speed Studies
 Spot Speed Studies
• Conducted for a point (spot) or a short
segment (<1,000 ft)
• Carried out to find out the choice of drivers
for speeds in free flow conditions, why??
• Not conducted when volumes are in excess of
750 to 1,000 veh/h/ln on freeways or 500
veh/h/ln on other types of uninterrupted flow
facilities.
 Uses of Spot Speed Data
• Establishing the effectiveness of new or existing speed
limits or enforcement practices.
• Specific design applications in determining appropriate
sight distances, relationships between speed and
highway alignment, and speed performance with
respect to steepness and length of grades.
• Specific control applications for the timing of "yellow“
and "all red" intervals for traffic signals, proper
placement of signs, and development of appropriate
signal progressions.
• Investigation of high-accident locations at which speed
is suspected to be a contributing cause to the accident
experience.
 Speed Definitions of Interest
• Average or time mean speed: the average speed of all
vehicles passing the study location during the period of the
study, mi/h.
• Standard deviation: in simplistic terms, the standard
deviation of speeds is the average difference between
individual observed speeds and the average speed during
the period of the study.
• 85th percentile speed: the speed below which 85% of the
observed vehicles travel, mi/h.
• Median speed: the speed that equally divides the
distribution of spot speeds; 50% of observed speeds are
higher than the median; 50% of observed speeds are lower
than the median, mi/h.
• Pace: a 10 mi/h increment in speeds that encompasses the
highest proportion of observed speeds (as compared with
any other 10 mi/h increment).
 Speed Data Collection
• Much speed data are collected
using permanent detector
locations. Loop detectors are the
most common form used in these
cases.
• Where no permanent detectors are •
in place at the desired study
location, most such measurements,
however, are made using hand-
held or vehicle-mounted radar
“guns” or detectors.
• Individual observed speeds will be
in the form of a distribution that will
have to be mathematically
described, individual speeds are
arranged in the form of frequencies
of observations within defined
speed groups.
This is done as it will facilitate the
statistical analyses and
determinations that will be
extracted from the data.
 Spot Speed Sample Data Analysis
• Typical breadth of speed group of 2 mi/hr,
smaller ranges can be used with larger dataset
• More than 5 mi/hr should never be used
Frequency Distribution Table
 Common Descriptive Statistics
• Central tendency: measures that describe the
approximate middle or center of the
distribution.
• Dispersion: measures that describe the extent
to which data spread around the center of the
distribution.
 Central Tendency: Mean

Central Tendency: Median

• Divides the observations in to equal halves
• Not affected by magnitude like mean
• Therefore, 50% of all observed speeds should be less
than the median.
• This can be estimated using cumulative frequency
distribution curve plots.
 Central Tendency: Pace
• A 10-mi/h increment in speeds that encompasses the
highest proportion of observed speeds (as compared
with any other 10-mi/h increment).
• A 10-mi/h template is scaled from the horizontal axis.
Keeping this template horizontal, place an end on the
lower left side of the curve and move slowly along the
curve. When the right side of the template intersects
the right side of the curve, the pace has been located.
This procedure identifies the 10-mi/h increment that
intersects the peak of the curve; this contains the most
area and, therefore, the highest percentage of vehicles.
 Central Tendency: Mode
• The mode is defined as the single value of
speed that is most likely to occur
• Peak of the probability distribution curve
 Dispersion: Standard Deviation
• In simplistic terms, the standard deviation of
speeds is the average difference between
observed speeds and the time mean speed
during the period of the study.
• For discrete values:
• For continuous data:
 % of Vehicles Within Pace
• The smaller the percentage of vehicles traveling within the
pace, the greater the degree of dispersion in the
distribution.
• The pace range speed is found from frequency distribution
curve as explained earlier
• Then the corresponding y-coordinates of these values from
cumulative distribution curve is determined

Even though speeds between 32 and 66 mi/h were observed in

this study, over 60% of the vehicles traveled at speeds between
42.5 and 52.5 mi/h. This represents expected traffic behavior with
a standard deviation of approximately 5 mi/h,
 Dispersion: 85th and 15th Percentile
Speeds
• These give a general description of the high and low
speeds observed by most reasonable drivers
• As these values get closer to the mean, less dispersion
exists and the stronger the central tendency of the
distribution becomes.
• Found from cumulative distribution curve at y-
coordinate of 15 and 85
• P85 = 54.7 mi/h
• P15 = 42.42 mi/h
• Can be used to calculate an approximate value for
standard deviation
 Using the Normal Distribution in the
Analysis of Spot Speed Data
• Most speed distributions tend to be
statistically normal (i.e., they can be
reasonably represented by a normal
distribution).
• standard notation x=N[40,25] signifies that
the variable “x" is normally distributed with a
mean of “40“ and a variance of "25“.
• standard normal distribution, where z: N[0,1 ]
 Precision and Confidence Interval
• The standard deviation of a distribution of
sample means with a constant group size of n

• Assuming a normal distribution, it is known

that 95% of all values lie between the mean ±
1.96 standard deviations; 99.7% of all values
lie between the mean ±3.00 standard
deviations.
 Required Sample Size
• Given that the precision or tolerance (e) of the
estimate is the ± range around mean:
 Sample Size Problem
• Consider the following problem: How many speeds must be
collected to determine the true mean speed of the
underlying distribution to within ±1.0 mi/h with 95%
confidence? How do the results change if the tolerance is
changed to ±0.5 mi/h and the confidence level to 99.7%?
 Before and After Spot Speed Studies
• The difference between two sample means follows normal distribution if
– Both are coming from the same distribution
– Sample size is 30 or more for both cases
• Two important questions:
– Was the difference in mean speeds significant?
– If yes, then was the desired reduction achieved or not?
• Found through statistical tests

• If Prob (z < Zd ) > 0.95, the observed reduction in speeds is statistically

significant.
• If Prob (z < Zd) < 0.95, the observed reduction in speeds is not
statistically significant.
 Example
• Consider the following results of a before-and-
after spot speed study conducted to evaluate
the effectiveness of a new speed limit
intended to reduce the average speed at the
location to 60 mi/h:
 Problem 10.2
• A before-and after speed study was conducted
to determine the effectiveness of a series of
rumble strips installed approaching a toll plaza
to reduce approach speeds to 40 mi/h.
Check for Normalcy: Chi-Square
 𝑥−𝑢
𝑧=
𝜎
`
• Degrees of freedom = number of groups – 3

• This table yields probability of equal or higher so

the decision criteria for 5% confidence interval.
• For the given case it is higher than 5% so the
difference between the given distribution is not
statistically significant than chi-square hence it
follows normal distribution
 Example 10.1
• Consider the spot speed data here, collected
at a rural highway site under conditions of
uncongested flow:
Speed group Number of vehicles
15-20 0
20-25 4
25-30 9
30-35 18
35-40 35
40-45 42
45-50 32
50-55 20
55-60 9
60-65 0
 10.1(a)
• Plot the frequency and cumulative frequency
curves for this data.
Speed group Number of Frequency (%) Cumulative
(mph) vehicles frequency(%)
15-20 0 0.00 0.00
20-25 4 2.37 2.37
25-30 9 5.33 7.69
30-35 18 10.65 18.34
35-40 35 20.71 39.05
40-45 42 24.85 63.91
45-50 32 18.93 82.84
50-55 20 11.83 94.67
55-60 9 5.33 100.00
60-65 0 0.00 100.00
 FREQUENCY CURVE
30.00

25.00

20.00

Frequency (%)
15.00

10.00

5.00

0.00
20 30 40 50 60 70
-5.00
Middle Speed (mph)

CUMULATIVE FREQUENCY CURVE

120.00
Cumulative frequency (%)

100.00

80.00

60.00

40.00

20.00

0.00
20 25 30 35 40 45 50 55 60 65
Upper speed limit (mph)
 10.1 (b)
• Determine the median speed, the modal
speed, the pace, and the percentage of
vehicles in the pace from the curves.
Cumulative frequency curve
120.00
Cumulative frequency (%)

100.00
80.00
60.00
40.00
20.00 Median = 42.5mph
0.00
20 25 30 35 40 45 50 55 60 65
Upper speed limit (mph)
 Frequency curve
30.00

25.00

20.00
Frequency (%)

15.00

10.00

5.00

0.00
20 30 40 50 60 70
-5.00
Middle Speed (mph) Cumulative frequency curve
120.00

Cumulative frequency (%)

100.00
80.00
60.00
40.00
20.00
0.00
20 25 30 35 40 45 50 55 60 65
Upper speed limit (mph)

• Pace = 37.5-47.5 mph, % vehicles within pace

= 78 – 22% = 56%
(c) Compute the mean and standard
deviation of the speed distribution.

Speed group Number of vehicles

15-20 0
20-25 4
25-30 9
30-35 18
35-40 35
40-45 42
45-50 32
50-55 20
55-60 9
60-65 0
 10.1(d)
• What are the confidence bounds of the
estimate of the true mean speed of the
distribution with 95% confidence? With 99.7%
confidence?
𝜇 = 𝑥 ± 1.96𝐸 𝑠
𝐸=
𝑛
𝜇 = 𝑥 ± 3𝐸
 10.1(e)
• Based on the results of this study, a second is
to be conducted to achieve a tolerance of ±
0.8 mi/h with 95% confidence. What sample
size is required?
 10.1(f)
• Can this data be adequately described as
"normal?"
speed group (mi/h)
observed
frequency upper Prob for Theoretical Combined Chi-
high speed low speed n limit Zd Prob z<zd the group f Combined f n square
∞ 55 9 ∞ 1 0.0568 9.5992 9.5992 9 0.037403
55 50 20 1.59556 0.9432 0.1354 22.8826 22.8826 20 0.363131
50 45 32 0.86806 0.8078 0.1672 28.2568 28.2568 32 0.495865
45 40 42 0.36251 0.6406 0.2393 40.4417 40.4417 42 0.060044
40 35 35 -0.254 0.4013 0.2091 35.3379 35.3379 35 0.003231
35 30 18 -0.8705 0.1922 0.1241 20.9729 20.9729 18 0.421407
30 25 9 -1.487 0.0681 0.0493 8.3317 11.5089 13 0.193188
25 -∞ 4 -2.1036 0.0188 0.0188 3.1772

Total Chi square = 1.57 Probability between 90 and 75%. It is a normal distribution