ESTIMATING RIGHT TURNS ON RED
AT INTERSECTIONS WITH PEDESTRIAN ACTIVITY
by
D. Holly Miller
B.S., University of Portland, 1993
A thesis submitted to the
University of Colorado at Denver
In partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering
2002
i
t_______________!;
This thesis for the Master of Science
degree by
D. Holly Miller
has been approved
by
Bruce Janson
Sarosh Khan
Miller, D. Holly (M.S., Civil Engineering)
Estimating Right Turns on Red at Intersections with Pedestrian Activity
Thesis directed by Associate Professor Bruce N. Janson
ABSTRACT
The number of vehicles turning right during the red phase at a signalized
intersection is rarely collected during field observations. This absence of data
requires the analyst to estimate the number of right turn on red vehicles using
a rule of thumb or assume the worstcase scenario of zero, in order to
conduct intersection capacity analyses. In either case, the estimates made
often do not reflect the actual right turn on red volumes.
This study attempts to provide a more refined methodology for estimating the
number of right turns on red as they relate to opposing through volumes,
pedestrian volumes, green time and the number of right turning vehicles.
Thirty 15minute data sets were gathered at downtown Denver intersections.
Multiple variable regression is used to relate the influence of each variable on
the right turn on red volume.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
Signed
Bruce N. Janson
hi
CONTENTS
Tables..................................................................vi
Figures................................................................vii
Chapter
1. Introduction..........................................................1
2. Problem Statement.....................................................2
3. Other Research...................................................... 3
3.1 Article by Luh and Lu (1990).........................................3
3.2 Article by Virkler and Ramana (1995)................................4
3.3 Article by AbuLebdeh et. al. (1997)................................4
3.4 Article by Tarko (2001).............................................5
3.5 Highway Capacity Manual (TRB, 2000).................................6
4. Software..............................................................8
4.1 Highway Capacity Software (McTrans, 2000)............................8
4.2 Signal 2000 (Strong Concepts, 2000).................................8
4.3 Synchro (Trafficware, 2001).........................................9
5. Data Collection......................................................10
6. Data Analysis........................................................15
6.1 Correlation.........................................................15
6.2 Regression.........................................................18
IV
6.2.1 Model 1...........................................................18
6.2.2 Model 2...........................................................19
6.2.3 Model 3...........................................................20
6.2.4 Model 4...........................................................21
6.2.5 Model 5...........................................................21
6.2.6 Model 6...........................................................22
6.2.7 Model 7...........................................................23
6.2.8 Model 8...........................................................23
7. Comparison of RTOR Model Results.....................................26
7.1 AbuLebdeh et. al. (1997) and 25% Estimate Models.................26
7.1.1 AbuLebdeh et. al. (1997) Model...................................26
7.1.2 25% Rule of Thumb.................................................27
7.2 Comparison of Developed RTOR Models...............................30
7.2.1 Model 1...........................................................34
7.2.2 Model 2...........................................................35
7.2.3 Model 3...........................................................36
7.2.4 Model 4...........................................................37
7.2.5 Model 5...........................................................38
7.2.6 Model 6...........................................................39
7.2.7 Model 7...........................................................40
v
7.2.8 Model 8
41
8. Summary and Conclusion...........................................42
Appendix
A. Data Set Figures.................................................45
B. Regression Results...............................................76
References..........................................................85
VI
TABLES
Table
1. DataSets.......................................................14
2. Correlation....................................................17
3. Comparison of Other Models....................................29
4. Comparison of Results........................................32
vii
FIGURES
Figure
1. AbuLebdeh et. al. (1997) Results..................................27
2. 25% of Total Right Turn Volume Results..............................28
3. Summary of Error in Predictions.....................................31
4. Model 1 Results...................................................34
5. Model 2 Results...................................................35
6. Model 3 Results...................................................36
7. Model 4 Results...................................................37
8. Model 5 Results...................................................38
9. Model 6 Results...................................................39
10. Model 7 Results...................................................40
11. Model 8 Results...................................................41
VIII
1. Introduction
This study provides a methodology for estimating the volume of rightturns on
red (RTOR) using data typically gathered in the field including right turn
volumes, lane configurations, opposing traffic volumes, pedestrian volumes
and green time to cycle length ratios (g/C). The study also shows what
influence pedestrian activity has on RTOR volumes.
1
2. Problem Statement
The number of vehicles turning during the red phase at a signalized
intersection is rarely collected during field observations. This absence of data
requires the analyst to estimate the volume using a rule of thumb or assume
the worstcase scenario of zero, in order to conduct intersection capacity
analysis. The hypotheses explored for this thesis is twofold:
1) RTOR volumes can be estimated using data typically collected for traffic
studies and
2) Pedestrian activity has a significant impact on RTOR volumes.
2
3. Other Research
A number of articles have been published describing possible methods for
dealing with the RTOR. Some explore methods for computing the capacity of
the RTOR while others describe the affect the RTOR has on intersection
safety. This section summarizes the research found that is relevant to this
thesis.
3.1 Article by Luh and Lu (1990)
In an article describing a methodology to compute the capacity of right turns
on red, Luh and Lu (1990) propose a procedure that involves modifying the
capacity computation for stopsigncontrolled right turns. The methodology
requires collecting some additional data from the field. The procedure is
derived from Chapter 10 Unsignalized Intersections in the Highway Capacity
Manual (TRB, 1985 ) and has not been validated with empirical data. Luh
and Lu (1990) deal specifically with the capacity of the RTOR movement but
do not provide a method for estimating the RTOR volume.
3
3.2 Article by Virkler and Ramana (1995)
Transportation Research Record 1484 includes an article by Virkler and
Ramana (1995) that describes two methods for estimating the RTOR
capacity. The first is based on determining the number of cross street left
turns made during a protected phase (RTOR can be made without conflict
during this cross street left turn phase). The second is based on the
unsignalized right turn capacity methodology described in the 1994 Highway
Capacity Manual (TRB, 1994). The article states that basing the RTOR
capacity on the cross street left turn volume is simple and conservative and
that basing the capacity on the unsignalized methodology yields a higher
capacity. It concludes that either methodology is better than assuming zero
right turns on red. Like Luh and Lu (1990), the methods described by Virkler
and Ramana (1995) deal strictly with the potential capacity of the RTOR
movement but do not provide a method for estimating the actual RTOR
volume.
3.3 Article by AbuLebdeh et. al. (1997)
Abu Lebdeh et.al. (1997) describes a method for estimating RTOR volumes
based on the following four factors: Right turn volume, conflicting volume,
4
green time divided by the cycle length and the type of conflicting traffic, which
accounts for being from the crossstreet approach or the opposing approach.
Abu Lebdeh et.al. (1997) found that although a simple model using only the
turning volume can be used, there are other factors that significantly influence
RTOR volumes.
In addition, Abu Lebdeh et.al. (1997) compares delay computations with and
without the right turn reduction for the intersections studied based on the
methodology suggested in the Highway Capacity Manual (TRB, 1994). The
effect of pedestrian traffic and the number of right turn lanes are not
accounted for in the model. The estimated RTOR volume, using the model
suggested in this article, will be compared to the model developed in this
thesis and the actual number of RTOR movements observed in the field.
3.4 Article by Tarko (2001)
Tarko (2001), presents a model for predicting the RTOR volumes at isolated
and coordinated signals. The three factors that were assumed to have an
effect on the RTOR volume are (1) the number of rightturners scheduled to
arrive during the red signal (2) the level of impedance from the queue on the
5
same approach, and (3) the level of impedance from traffic flow into which
right turns attempt to merge or cross. CORSIM was used to generate
benchmark data and assumed to replicate a realworld environment. No field
data was collected to verify the results of the model.
3.5 Highway Capacity Manual (TRB, 2000)
The Highway Capacity Manual (TRB, 2000), states that when right turns on
red are permitted, the rightturn volume for analysis may be reduced by the
volume of rightturning vehicles moving on the red phase. At an existing
intersection, the number of right turns on red should be determine by field
observation. In the absence of field data, it is preferable in most cases to
utilize the right turn volumes directly without a reduction for right turns on red
except when an exclusive right turn lane movement runs concurrent with a
protected left turn phase from the cross street. In this case, total right turn
volume for analysis can be reduced by the number of cross street left turns
during the protected phase.
Pedestrian activity is not specifically addressed in the right turns on red
section of the HCM 2000. Instead, the effects of pedestrian activity are
6
accounted for as part of the right turn factor which determines the total
capacity of the right turn movement during the green phase. However, Luh
and Lu (1990) and Virkler and Ramana (1995) both examine techniques for
estimating RTOR volumes utilizing the Highway Capacity Manual (TRB,
1985) (TRB, 1994) unsignalized intersection methodology which does take
into account the effect of pedestrian activity.
7
4. Software
A number of different software programs are used to calculate capacity of
signalized intersections. Most programs emulate the methodology described
in the Highway Capacity Manual (TRB, 2000). This section describes how
each of the software programs handles RTOR volumes and the methodology
suggested in the Highway Capacity Manual.
4.1 Highway Capacity Software (McTrans, 2000)
Highway Capacity Software emulates the methodology defined in the
Highway Capacity Manual (TRB, 2000). The help files provided with the
Highway Capacity Software state that the number of vehicles that execute
right turns on red are entered in vehicles per hour using the guidelines
discussed on page 169 of the Highway Capacity Manual (TRB, 2000).
4.2 Signal 2000 (Strong Concepts, 2000)
Signal 2000 (Strong Concepts, 2000) software also emulates methodologies
defined in the Highway Capacity Manual (TRB, 2000). The documentation
states that the RTOR volume entered will be used to reduce the right turn
8
volume before any other adjustments are made, with the limitation that the
right turn volume will never be reduced below one vehicle per hour.
4.3 Synchro (Trafficware, 2001)
Synchro (Trafficware, 2001) calculates a saturation flow rate for the RTOR
and Left Turns on Red for crossing oneway streets. The calculation is based
on the signal timing, the volumes of the approach, and the volumes of any
merging approaches. The methodology is taken from the Highway Capacity
Manual Unsignalized Intersection Methodology chapter. The maximum flow
rate with zero conflicting vehicles assumes one vehicle every 3.3 seconds or
1,091 vehicles per hour. Conflicting pedestrian volumes are added into the
merging vehicle volumes when calculating the saturation flow rate.
9
5. Data Collection
Thirty data sets were gathered at various downtown Denver intersections
between 1998 and 2001. Downtown intersections were chosen to ensure that
the influence of pedestrian activity was taken into account. With a few
exceptions, RTOR are permitted in downtown. Only intersections where
RTOR activity is permitted were studied. Data sets were recorded in 15
minute periods. Observations were made on a variety of days with varying
levels of traffic and pedestrian activity including a day when pedestrian
activity was exceptionally high with people heading to a Rockies baseball
game.
In the heart of downtown Denver, many intersections provide an all
pedestrian phase. During the allpedestrian phase, pedestrians can cross the
street diagonally as well as cross any of the four legs of the intersection.
Motorists are allowed to make a RTOR during the pedestrian phase but must
yield to pedestrians. Pedestrians are not permitted to cross the path of a right
turning vehicle during the right turning vehicles green phase. All pedestrian
activity that conflicts with RTOR activity, including the activity during the all
pedestrian phase is included in the data sets.
10
The majority of data sets were gathered at the intersection of two oneway
streets. This allowed a few leftturn on red data sets to be gathered in
addition to RTOR data sets. Because left turn on red activity is similar to
RTOR activity at the intersection of two oneway streets these data sets are
included in the analysis and are referred to as right turns.
The following information was recorded for each data set:
T = Start time of the 15minute interval
TRTV = Total right turn volume
RTOR = Right turn on red volume
Opp Vol = Opposing through volume (the through volume
approaching from the leg to the left of the right turning vehicles)
g/C ratio = green time/cycle length ratio
RT Lane = number of right turn lanes (0.5 for a shared through/tum
lane, 1 for a single exclusive turn lane and 1.5 for a single exclusive
lane plus a shared turn/through)
Peds = Pedestrian volume that conflicts with right turn on red
activity
11
Opp Lane = Number of opposing through lanes including
shared/through lanes
Table 1 lists each of the data sets and information recorded. Figures for each
of the data sets are included in Appendix A. As shown, the RTOR volume
varied between one and 29 vehicles with an average of nine vehicles per 15
minute period. The total right turn volume varied between seven and 63 with
an average of 31 per 15minute period. The conflicting pedestrian volume
varied between two and 243 with an average of 59 per 15minute period.
Other factors that influence the RTOR volume include:
o Arrival pattern (number of turning vehicles that arrive during the
red phase vs. the green phase)
o Opposing volume lane utilization factors (the number of through
vehicles using the lane that the right turn is entering vs. the
other lanes)
o Right turn lane utilization factors i.e. the number of through
vehicles in a shared through/right turn lane
o Left turn shadow (the period of cycle when the opposing left
turns shadows the right turning vehicles) this is only applicable
to twoway streets with a protected left turn phase
12
o Pedestrian grouping while collecting data, it was noted that a
platoon of pedestrians (ie a family) did not appear to impede
traffic as greatly as a stream of individual pedestrians
Although these factors influence the RTOR volumes, this information is not
typically collected in the field and would therefore require the analyst to make
assumptions about them. For this reason, these factors were not used to
create the models developed in this thesis.
13
TABLE 1. DATASETS
Opp Opp g/C RT
Data Set RTOR Vol TRTV Peds Lane ratio Lane
1 Larimer/17th 9 156 36 143 4 0.33 1.5
2 Araphahoe/17th 11 96 21 36 4 0.33 1.5
3 Market/17th 9 75 45 169 3 0.50 0.5
4 Lawrence/17th 16 125 44 53 4 0.33 1.5
5 Larimer/17th 4 169 24 167 4 0.33 1.5
6 Larimer/18th 16 261 47 98 4 0.33 1.5
7 Lawrence/18th 9 179 63 55 4 0.33 1.5
8 Larimer/17th 6 175 27 243 4 0.33 1.5
9 Curtis/17th 17 173 35 83 4 0.33 1.5
10 Arapahoe/17th 11 231 31 110 4 0.33 1.5
11 Arapahoe/17th 5 142 34 76 4 0.33 1.5
12 Lawrence/15th 20 94 58 40 4 0.33 1.5
13 Arapahoe/15th 21 183 58 34 4 0.33 1.5
14 Curtis/15th 12 37 53 58 2 0.33 1.5
1520th/Champa 5 72 14 2 3 0.50 1.5
16 L20th/Champa 1 49 9 8 3 0.50 1.0
17 17th/Wazee 7 36 58 22 2 0.50 0.5
18 17th/Wazee 1 29 7 21 2 0.50 0.5
19 18th/Market 1 64 24 17 3 0.50 0.5
20 L18th/Market 1 54 15 13 4 0.50 0.5
21 19th/Blake 4 16 7 69 2 0.50 0.5
22 19th/Market 9 30 15 94 3 0.50 1.0
23 19th/Lawrence 14 20 31 19 4 0.33 1.5
24 Stout/15th 10 15 55 32 2 0.33 1.0
25 Champ/15th 29 96 38 36 4 0.33 1.5
26 Sout/19th 12 62 14 6 3 0.33 1.0
27 L Stout/19th 1 52 8 17 2 0.33 1.5
28 Stout/18th 9 49 15 13 2 0.33 1.0
29 Curtis/18th 7 22 24 12 3 0.33 1.5
30 Champa/18th 4 105 8 11 3 0.33 1.5
Low 1 15 7 2 2 0.33 0.5
High 29 261 63 243 4 0.5 1.5
Average 9 96 31 59 3.3 0.4 1.2
L = Left turn data sets.6.0
14
6. Data Analysis
6.1 Correlation
Correlation analysis is used to compare the RTOR volumes to each of the
variables collected. The intent is to identify those variables that are highly
correlated with the RTOR volume to develop a multi variable regression
model. In addition, variables highly correlated with each other or that have a
low correlation with RTOR volumes are not be used to develop the RTOR
estimation model.
Table 2 lists the correlation numbers for the data set. As shown, the RTOR
variable has absolute correlation values greater than 0.4 with TRTV (the total
right turn volume), Opp Lane (the number of opposing through lanes), the g/C
ratio, and RT Lanes (the number of right turn lanes). The number of right turn
lanes variable appears to be highly correlated with the number of opposing
through lanes and the g/C variable. This is likely due to the similar
geometries at many of the intersections and the constant 75second cycle
length in all the data sets and the number turn lanes varying between 0.5 and
1.5 but are not logically related to each other. Because of this, all three
variables will be used to develop regression models. The correlation between
15
the RTOR volume and the number of conflicting pedestrians was low (0.03).
However, because part of the intent of this thesis is to determine the influence
of pedestrians on the RTOR volume, this variable will be also be included in
developing some of the regression models.
A second set of correlation data was developed using a 1/0 variable for both
the pedestrian phase and the number of right turn lanes. One was entered
for data sets with a pedestrian phase and zero for those without a pedestrian
phase. Similarly, one was entered for data sets with a separate right turn
lane and zero was entered for those with only a shared right turn lane. The
correlation values were similar to those found in Table 2. Therefore the
original data set is used to develop the regression models.
16
TABLE 2.CORRELATION
RTOR Opp Vol TRTV Peds Opp Lane g/C ratio RT Lane
RTOR 1.00 l*1* ^ ifclilllS ^ lÂ§5Â§$ I'h te:~: O ;^llp^lH i
Opp Vol 0.32 1.00 I WH!
TRTV 0.59 0.35 1.00 Mp : O' .*,
Peds 0.03 0.53 0.18 1.00 * siCi. ;0: t:^L:^:^:il!!iii'''!^ '? : !!:to
Opp Lane 0.43 0.70 0.27 0.36 1.00 ft1 4' Iplllll appStl
g/C ratio 0.51 0.46 0.33 0.13 0.39 1.00 [.,#. '$:< E i:
RT Lane 0.45 0.54 0.24 0.16 0.54 0.80 1.00
6.2 Regression
Multiple variable regression is used to relate the effect that each variable has
on the RTOR volumes. The variables chosen are Opp Lane (the opposing
number of lanes), TRTV (total right turning volume), Peds (conflicting
pedestrian volume), g/C ratio (green time to cycle length ratio) and RT Lane
(the number of turn lanes). Eight models are developed run using a
combination of one or more of these five variables. Detailed results for each
of the eight regression models are included in Appendix B.
6.2.1 Model 1
RTOR = 0 + (0.28 TRTV)
R2 = 0.31
t(intercept) = na; t(TRTV) = 9.81
The first model, a single variable linear regression model, estimates the
RTOR volume using TRTV (the total right turn volume) only. The intercept is
forced to zero since there should be no RTOR vehicles if the total right turn
volume is zero. The R squared for this model is 0.31. The regression model
yields a coefficient of 0.28 for the total turn volume with an intercept of zero.
This indicates that the model always yields a RTOR volume of 28% of the
total turn volume. The field data collected shows that the RTOR volume
18
varied between 4% and 86% of the total turn volume with an average of 33%.
Although the model yields results similar to the average RTOR volume
observed, this indicates that there are other factors, in addition to the total
right turn volume, that influence RTOR volumes. For comparison, Abu
Lebdeh et al. (1997) found that RTOR volumes accounted for between 11%
and 77% of the total right turn volume with an average of 38%.
6.2.2 Model 2
RTOR =2.99+.23 (TRTV) 0.01 (Peds)
R2 = 0.35
t(intercept) = 1.36; t(TRTV) = 3.82; t(Peds) = 0.52
The second model utilizes TRTV (the total right turn volume) and Peds
(conflicting pedestrian volume) variables to develop a regression model. The
R squared for this model is 0.35, which indicates that it should predict RTOR
volumes slightly better than Model 1. The TRTV coefficient is positive and the
Peds coefficient is negative. This matches the expectation that as the
number of conflicting pedestrians increases, the RTOR volume decreases.
The t statistic for the TRTV is higher than two (3.82) indicating that it is
significant. However the t statistic for the Peds variable is low indicating that
it may not be significant to the results of the model.
19
6.2.3 Model 3
RTOR = 5.06 + (2.98 Opp Lane) + (0.20 TRTV) (0.02 Peds)
R2 = 0.46
t(intercept) = 1.26; t(Opp Lane) = 2.31; t(TRTV) = 3.51; t(Peds) = 1.29
The third model utilizes Opp Lane (the number of opposing lanes), TRTV (the
total right turn volume) and Peds (conflicting pedestrian volume) to develop a
regression model. The R squared for this model is 0.46 which indicates that
is should predict the RTOR volumes better than Models 1 and 2. The
coefficients for both Opp Lane and the TRTV are positive indicating that as
each of these increases, the predicted RTOR volume increase. This agrees
with the expectation that as the number of opposing lanes increases, the
RTOR volume increases. The t statistics for both the number of opposing
lanes variable and the total turn volume variable are above two indicating that
they are significant to the model. However, the Peds variable is less than two
indicating that it may not be significant to the results of the model.
20
6.2.4 Model 4
RTOR = 16.27 + (30.85 *g/C) + (0.18 TRTV) (0.03 Peds)
R2 = 0.46
t(intercept) = 2.7; t(g/C) = 2.35; t(TRTV)= 3.18; t(Peds)= 0.75
The fourth model utilizes the g/C ratio, TRTV (the total right turn volume) and
Peds (the conflicting pedestrian volume) to develop a regression model. The
R squared for this model is 0.46 which indicates that is should predict RTOR
volumes about the same as Model 3. The coefficient for the g/C ratio is
negative indicating that the RTOR volume will decrease as the green time for
the right turn increases and the amount of red time decreases. The t
statistics for both the g/C ratio and the TRTV variable are above two
indicating that they are significant to the model. However, like the other
models, the Peds variable is less than two indicating that it may not be
significant to the results of the model.
6.2.5 Model 5
RTOR = 2.84 + (5.73 RT Lane) + (0.19 TRTV) (0.01 Peds)
R2 = 0.46
t(intercept) = 0.88; t(RT Lane) = 2.33; t(TRTV) = 3.52; t(Peds) = 0.84
The fifth model utilizes RT Lane (the number of right turn lanes), TRTV (total
right turn volume) and Peds (conflicting pedestrian volume) to develop a
21
regression model. The R squared for this model is 0.46 which indicates that
is should predict RTOR volumes with similar results as Models 3 and 4. The
coefficient for the RT Lane variable is positive indicating that more right turn
lanes increases the RTOR vehicles, as expected. The t statistics for both RT
Lane and TRTV variables are above two indicating that they are significant to
the model. However, the Peds variable t statistic is less than two indicating
that it may not be significant to the results of the model.
6.2.6 Model 6
RTOR = 6.16+ (0.41 RT Lane) + (0.17 TRTV) (0.02 Peds) (22.00 g/C) + (2.81 Opp Lane)
R2 = 0.52
t(lntercept) = 0.49; t(RT Lane) = 0.09; t(TRTV) = 3.00; t(Peds) = 1.26; t(g/C) = 1.04; t(Opp Lane) =
1.48
The sixth model utilizes all six variables to develop a regression model. The
R squared for this model was 0.52 which indicates that it should predict the
RTOR volumes better than the other models developed. However, the t
statistics for all variables except the TRTV are quite low indicating that they
may not be significant to the model.
22
6.2.7 Model 7
RTOR = 4.41+ (0.19 TRTV) + (2.43 Opp Lane)
R2 = 0.43
t (Intercept) = 1.09; t(TRTV) = 3.36; t(Opp Lane) = 1.97
The seventh model utilizes the two variables with the highest t statistics from
Model 6 without the conflicting pedestrian volume. This includes TRTV (the
total right turn volume) and Opp Lane (the number of opposing lanes). The R
squared for this model is 0.43 indicating that it predicts RTOR volumes with a
similar accuracy to a number of the models previously developed. Like the
other models developed, the TRTV variable t statistic is greater than two.
The Opp Lane variable has a t statistic slightly less than two.
6.2.8 Model 8
To develop Model 8, intersections with g/C ratios of 0.33 and 0.50 are
separated. There are 21 intersections with a g/C ratio of 0.33. All of these
have a 25 second allpedestrian phase. There were nine intersections with a
g/C ratio of 0.50. These have a simple twophase signal without a separate
pedestrian phase.
23
The model developed for the intersections with a g/C of 0.33 yields a similar
R squared value as the previous models and the t statistic for pedestrian
activity is less than two, indicating that it is not significant to the model.
A number of models were developed for the data sets with a 0.50 g/C ratio.
Using Opp Vol (the Opposing volume), TRTV (total right turn volume), Peds
(conflicting pedestrian volume) and RT Lane (number of right turn lanes)
yields the highest R squared value. The model shown below for these nine
intersections yields an exceptionally high R squared value of 0.96. In this
model the t statistics, for each of the variables used, is above two indicating
that they are significant to the model.
RTOR = 1.87 (0.06*Opp Vol) + (0.12*TRTV) + (0.04*Peds) + (6.14* RT Lane)
R2 = 0.96
t(intercept) = 1.62; t(Opp Vol) =3.24; t(TRTV) = 5.21; t(Peds) = 6.29; t(RT Lanes) = 5.29
Unfortunately, there are a few concerns that are raised with the use of this
Model.
The data set is small without a large variation in the RTOR volume.
Gathering additional data sets could be used to verify the validity.
24
The g/C variable is not used in Model 8 because it is a constant
0.50 for all the data sets. It is possible that estimates generated
using Model 8 could factor the resulting RTOR by the appropriate
g/C factor i.e if the g/C is 0.25 the red time would be 0.75 (50%
higher than what the model was based on). This would provide
50% more time for RTOR activity to occur. If the model estimates
10 RTOR vehicle, using the assumption that there would be 50%
more red time for RTOR activity would result in a RTOR volume of
15 vehicles. Gathering additional data sets could be used to verify
this estimate.
The model implies that the RTOR volume increases as the
conflicting pedestrian (Peds) volume increases based on the
positive sign (+0.04 Peds) in the equation. Intuitively however,
the RTOR volume should go down as the number of conflicting
pedestrians increases.
25
7. Comparison of RTOR Model Results
The following set of tables compare the results of the models developed in
this thesis, a model developed by AbuLebdeh et. al. (1997) and results using
a rule of thumb often used by transportation professionals to RTOR volumes
observed in the field for this project. The high, low and average predictions
are used to compare the accuracy of each of the models.
7.1 AbuLebdeh et. al. (1997)
and 25% Estimate Models
Table 3 applies the model developed by AbuLebdeh et. al. (1997) and a 25%
rule of thumb estimate to the data set and compares the results to the
observed RTOR volumes.
7.1.1 AbuLebdeh et. al. (1997) Model
As shown, the model developed by AbuLebdeh et. al. (1997) predicted a
high RTOR volume of 42 vehicles for data set #24. This is substantially
higher than the observed RTOR volume for this data set of only 10 vehicles.
This model predicts an average of 21 RTOR vehicles compared to the
26
observed average of nine with an average difference of 14 RTOR vehicles.
As shown, the model also makes a number of predictions below zero.
Figure 1. AbuLebdeh et. al. (1997) Results
7.1.2 25% Rule of Thumb
Using a rule of thumb that RTOR vehicles account for 25% of the total right
turn volume yields a maximum difference of 20 vehicles less than the RTOR
volume observed for data set 25 where 29 RTOR vehicles were observed.
On average, this method estimates eight RTOR vehicles with an average
difference of four RTOR vehicles.
27
Figure 2. 25% of Total Right Turn Volume Results
35
30
25
2 20
>
o 15
Â£
10
I Prediction
Observed
28
TABLE 3.
COMPARISON OF OTHER MODELS
29
7.2 Comparison of Developed RTOR Models
Table 4 (two parts) compares the results of each of the models developed
previously, to the observed RTOR volumes. It is interesting to note that each
of the models predicts an average RTOR volume of nine (with data sets 11
and 25 included), the same average as the observed RTOR volumes and,
only one negative number is predicted in all of the data sets. The results
show that two of the data sets (#11 and #25) account for the largest
differences between the predicted volume and the observed volume for seven
of the eight models. This may indicate that these two data sets are outliers.
For the purpose of this comparison, Table 4 includes the maximum difference
observed for these two data sets as well as the maximum differences
excluding them. Results presented in the following discussion do not include
data sets #11 and #25.
Figure 3 illustrates the range of predictions for each of the eight models. The
largest prediction higher and lower than the actual observed volume and the
average difference between observed and predicted for the entire data set
are shown on the figure. As shown, Model 8 yields the most accurate results
with the average difference between the observed and predicted models
30
falling just above zero and the highest and lowest differences both falling
close to the average. For all the other models, the average difference
between observed and predicted values falls between three and four.
Figure 3. Summary of Error in Predictions
Highest, Lowest and Average Difference
Predicted vs. Actual RTOR Volumes
10
5
0
5
10
1* X X X X X X
X
o t <3 lo A
jF jF jF jF jF jF jF jF
X Highest
Lowest
Average
31
TABLE 4. COMPARISON OF RESULTS
Model 1 Model 2 Model 3 Model 4
Data Set RTOR Observed Predicted Difference Predicted Difference Predicted Difference Predicted Difference
1 9 10 1 10 1 11 2 11 2
2 11 13 2 12 1 9 2 7 4
3 9 6 3 7 2 10 1 9 0
4 16 13 3 12 4 14 2 13 3
5 4 7 3 7 3 8 4 8 4
6 16 13 3 13 3 14 2 13 3
7 9 8 1 7 2 7 2 8 1
8 6 10 4 10 4 12 6 11 5
9 17 9 8 9 8 10 7 10 7
10 11 17 6 16 5 17 6 16 5
11 .  5 . 17 ' 12 . : '16 11 18 : .. 13. ' 16 ; . .11. 
12 20 15 5 14 6 10 10 15 5
13 21 18 3 17 4 18 3 17 4
14 12 10 2 10 2 12 0 11 1
15 5 4 1 6 1 7 2 3 2
16 1 3 2 5 4 5 4 2 1
17 7 17 10 16 9 12 5 11 4
18 1 2 1 4 3 2 1 2 1
19 1 7 6 8 7 8 7 5 4
20 1 4 3 6 5 10 9 3 2
21 4 2 2 4 0 1 3 1 3
22 9 4 5 5 4 5 4 2 7
23 14 9 5 10 4 13 1 12 2
24 10 16 6 15 5 11 1 16 6
25%tf o CM 7f7.11. 7 ..,71 e. m 7 11 . 18... 14 ' 7'1'5.\;, 13'iffif : 16
26 12 4 8 6 6 6 6 9 3
27 1 2 1 5 4 2 1 7 6
28 9 4 5 6 3 4 5 9 0
29 7 7 0 8 1 8 1 10 3
30 4 2 2 5 1 5 1 7 3
High" 21 18 10 17 9 18 9 17 6
Low 1 2 8 4 8 1 10 1 7
#11 5 17 12 16 11 18 13 16 11
#25 29 11 18 11 18 14 15 13 16
Average" 9 8 4 9 4 9 4 9 3
* Does not include data sets with an allpedestrian phase.
Does not include #11 or #25.
CO
CO
TABLE 4. COMPARISON OF RESULTS (CONT.)
Model 5 Model 6 Model 7 Model 8*
RTOR
Data Set Observed Predicted Difference Predicted Difference Predicted Difference Predicted Difference
1 9 11 2 11 2 12 3
2 11 7 4 6 5 9 2
3 9 9 0 11 2 11 2 9 0
4 16 14 2 15 1 14 2
5 4 8 4 9 5 10 6
6 16 14 2 14 2 14 2
7 9 8 1 7 2 17 8
8 6 12 6 12 6 10 4
9 17 10 7 11 6 12 5
10 11 17 6 17 6 11 0
11' .. 5 : I ..17 . ' 12 ,. 17,; 12 . 12' .A; r; ,
12 20 15 5 12 8 16
13 21 17 4 18 3 16 5
14 12 11 1 12 0 11 1
15 5 9 4 5 0 6 1 5 0
16 1 5 4 3 2 5 4 3 2
17 7 11 4 9 2 12 5 7 0
18 1 1 0 0 1 2 1 1 0
19 1 5 4 6 5 7 6 1 0
20 1 3 2 6 5 8 7 0 1
21 4 0 4 1 5 2 2 4 0
22 9 5 4 3 6 6 3 8 1
23 14 12 2 13 1 11 3
24 10 13 3 12 2 11 1
25, 29 13 16 .14' 15  13V> 16':
26 12 6 6 8 4 6 6
27 1 7 6 5 4 2 1
28 9 6 3 6 3 3 6
29 7 10 3 10 3 7 0
30 4 7 3 7 3 4 0
High** 21 17 6 18 6 17 8 9 2
Low** 1 0 7 1 8 2 6 0 1
#11 5 17 12 17 12 12 7 na na
#25 29 13 16 14 15 13 16 na na
Average** 9 9 3 9 3 9 3 4 0
* Does not include data sets with an allpedestrian phase.
** Does not include #11 or #25.
7.2.1 Model 1
This model predicts a maximum RTOR volume of 18 vehicles, a minimum of
two RTOR vehicles and an average of eight. This compares favorably to the
observed high, low and average volumes of 21,1 and 9, respectively. The
largest difference between predicted and observed volumes occurs for data
set 17 where the predicted volume is 17 and the observed volume is seven.
These results are similar to those found using the 25% of total right turn
volume rule of thumb.
Figure 4. Model 1 Results
25
Data Sets
34
7.2.2 Model 2
This model predicts a maximum RTOR volume of 17 vehicles, a minimum of
four and an average of nine. The model predicts the RTOR volume within an
average of four vehicles, 10 closer than the AbuLebdeh et.. al. (1997) model.
The model does not predict any negative values for the data set.
Figure 5. Model 2 Results
25
CO
Ui
20
3 15
O
>
o: 10
o
h
oc
Prediction
Observed
Data Sets
35
7.2.3 Model 3
Applying Model 3 yields a maximum RTOR volume of 18 vehicles, a minimum
RTOR volume of one vehicle and an average of nine. The maximum
difference is predicted for data set 12 where 10 fewer RTOR vehicles are
predicted than observed. The average difference is four RTOR vehicles. The
model did not predict any negative values and predicted the RTOR value on
average ten vehicles closer than the model developed by AbuLebdeh et.al.
(1997).
Figure 6. Model 3 Results
36
7.2.4 Model 4
Applying Model 4 yields a maximum RTOR volume of 17, a minimum RTOR
volume of one vehicle and an average of nine. The maximum difference
predicted is seven. The average difference is three RTOR vehicles,
somewhat better than the other three models. The model did not predict any
negative values and predicted the RTOR value on average eleven vehicles
closer than the model developed by AbuLebdeh et. al. (1997).
Figure 7. Model 4 Results
37
7.2.5 Model 5
Applying Model 5 yields a maximum RTOR volume of 17, a minimum RTOR
volume of zero and an average of nine. The maximum difference between
predicted and observed values is seven. The average difference is three
RTOR vehicles, similar to Model 4. The model did not predict any negative
values and predicted the RTOR value on average eleven vehicles closer than
the model developed by AbuLebdeh et. al. (1997).
Figure 8. Model 5 Results
38
7.2.6 Model 6
Applying Model 6 yields a maximum RTOR volume of 18, a minimum RTOR
volume of 1 and an average of nine. The maximum difference between
predicted and observed values is eight. The average difference is three
RTOR vehicles, similar to Model 4 and Model 5. The model does predict a
negative value.
Figure 9. Model 6 Results
24
Data Sets
39
7.2.7 Model 7
Applying Model 7 yields a maximum RTOR volume of 16, a minimum RTOR
volume of two and an average of nine. The maximum difference between
predicted and observed values is eight. The average difference is three
RTOR vehicles. The model does not predict any negative values.
Figure 10. Model 7 Results
40
7.2.8 Model 8
By removing the data sets collected at intersection with an allpedestrian
phase and applying Model 8 only to those intersection with a more typical
signal phasing, Model 8 predicts nearly the exact number of RTOR vehicles
for all nine data sets. The maximum difference between the actual RTOR
volume and the predicted volume is two vehicles and it predicts the actual
number of observed RTOR vehicles for six of the nine data sets. This model
clearly yields the most accurate results of the models developed but raises a
number of concerns as described earlier.
Figure 11. Model 8 Results
Data Sets
41
8. Summary and Conclusion
The number of vehicles turning right during the red phase at a signalized
intersection is rarely collected during field observations. This absence of data
requires the analyst to estimate the number of right turn on red vehicles using
a rule of thumb or assume the worstcase scenario of zero, in order to
conduct intersection capacity analyses. RTOR volumes are often estimated
using a rule of thumb such as 25% of the total right turn volume. However,
the RTOR volumes collected for this thesis varied between 4% and 86% of
the total right turn volumes indicating that there are other factors that
influence the RTOR volume. Estimates made often do not reflect the actual
right turn on red volumes.
This study attempts to provide a more refined methodology for estimating the
number of right turns on red as they relate to opposing through volumes,
number of opposing lanes, pedestrian volumes, green time to cycle length
ratio, the number of right turning vehicles and the number of right turn lanes.
Thirty 15minute data sets were gathered at downtown Denver intersections.
Multiple variable regression is used to relate the influence of each variable on
the number of turning vehicles. Eight different regression models are
42
developed in this thesis using a variety of information typically collected in the
field.
Intuitively, pedestrian activity has an impact on RTOR activity. However, only
one of the eight models developed in this thesis indicates that the conflicting
pedestrian variable has a significant influence on right turn on red predictions.
This may be due to the many other factors that influence traffic such as
vehicle arrival patterns, lane utilization factors, and the arrival patterns of
pedestrians.
Intersections with an allpedestrian phase do not yield as accurate of results
using the models developed in this thesis as those intersections without an all
pedestrian phase. The highest R squared value for the seven models that
include the allpedestrian data sets is Model 6 with an R squared of 0.52.
Model 6, (RTOR = 6.16+ (0.41 RT Lanes) + (0.17 TRTV) (0.02 Peds) 
(22.00 g/C) + (2.81 Opp Lane)), uses the number of right turn lanes, the
total right turn volume, conflicting pedestrian volume, g/C ratio and the
number of opposing lanes to estimate right turn on red volumes. However,
the t statistics are low for a number of the variables indicating that they may
not be significant.
43
When estimating RTOR volumes for intersections without an all pedestrian
phase, Model 8 ((RTOR = 1.87 (0.06*Opp Lane) + (0.12*TRTV) +
(0.04*Peds) + (6.14* RT Lane)) opposing volumes, the conflicting pedestrian
volume and the number of right turn lanes appear to have a significant
influence on the RTOR volume. This model yields an exceptionally good R
squared value of .96. However, this data set is small and additional data
should be collected to verify the results. Based on these results, it appears
that there is more potential for predicting RTOR activity at intersections
without an allpedestrian phase. The allpedestrian phase adds another
variable which does not allow the RTOR activity to be predicted with as much
accuracy as for those intersections without an allpedestrian phase.
44
Appendix
A. Data Set Figures
DATA SET: #1
Location: Larimer Street/17th Street
g/C ratio: 0.33
Date: April 21,1999
Time: 5:035:18 PM
No Scale
Legend
* Lane Configuration
XJUXX Total Right Turn Votune/Right Turn on Red Volune
XX Opposing Through Vdmv
Conflicting Pedestrians
46
47
DATA SET: #3
Location: Market Street/17th Street
g/C ratio: 0.50
Date: April 28,1999
Time: 5:105:25 PM
J
2*
75 >
No Scale
Lagand
Lanr Configuration
XX/XX Total Right Turn Vottew/Right Turn on Red Volune
XX Opposing Through Votive
X:I Conflicting Pedestrians
48
DATA SET: #4
Location: Lawrence Street/17th Street
g/C ratio: 0.33
Date: April 7, 1999
Time: 5:40 5:55 PM
J 17th STREET
No Scale
Legend
Lane Configuration
XX/XX o Total Right Turn Volunt/Right
XX b Opposing Through Volune
(jXj Conflicting Pedestrians
Ti^n on Red Volune
49
50
DATA SET: #6
Location: Larimer Street/18th Street
g/C ratio: 0.33
Date: June 8,1999
Time: 5:135:28 PM
47/16
51
52
53
DATA SET: #9
Location: Curtis Street/17th Street
g/C ratio: 0.33
Date: March 6, 2000
Time: 5:45 6:00 PM
J
173
No Scale
17th STREET
L
Legend
Lane Configuration
)0(/J0(.B Total Right Turn Volune/Right Turn on Red Volune
XX Opposing Through Volune
fcXil 8 Conflicting Pedestrians
54
55
56
DATASET: #12
Location: Lawrence Street/15th Street
g/C ratio: 0.33
Date: March 27, 2000
Time: 4:11 4:26 PM
M 11 J
I 94 I
Legend
Lone ConFlQurotion
XX/XX Totol Right Turn Volteie/Right Turn on Red Volune
xx Opposing Through Volme
[m Conflicting Pedestrians
57
DATA SET: #13
Location: Arapahoe Street/15th Street
g/C ratio: 0.33
Date: March 27, 2000
Time: 4:31 4:46 PM
58/21
Legend
* Lone Configuration
XX/XX e Tetol RrQht Turn Volunv/IBght Turn on Red Votune
XX Opposing Through Voiitfw
l'X;l Conflicting Pedestrians
58
DATA SET: #14
Location: Curtis Street/15th Street
g/C ratio: 0.33
Date: March 27, 2000
Time: 4:50 5:05 PM
vn r
37 1
53/12
No Scale
Legend
Lane Configuration
XX/XX Total Right Ti^n Volw/Right Turn on Red Volwne
XX Opposing Through Volune
^ Conflicting Pedestrians
59
DATA SET: #15
Location: Champa Street/20th Street
g/C ratio: 0.50
Date: July 5, 2001
Time: 6:30 6:45 PM
14/5
*M I l I
1 m I
Legend
Lane Configuration
XM(X Total Right Turn Volune/Right Turn on Red Volune
XX Opposing Through Volune
Conflicting Pedestrians
60
61
DATASET: #17
Location: Wazee Street/17th Street
g/C ratio: 0.50
Date: July 5, 2001
Time: 6:557:10 PM
J 17th STREET
U
36
58/7
No Scale
Legend
Lane Configuration
nuxx Total Right Turn Volm/Right Turn on Red Volune
XX Opposing Through Volini
Fx! Conflicting Pedestrians
62
DATA SET: #18
Location: Wazee Street/17th Street
g/C ratio: 0.50
Date: July 5, 2001
Time: 6:557:10 PM
$
I i T* I
29
No Scale
Legend
= Lane Configuration
XX/XX e Total Right Turn Volune/Right Turn on Red Volune
XX Opposing Through Volune
a Conflicting Pedestrians
63
DATASET: #19
Location: Market Street/18th Street
g/C ratio: 0.50
Date: July B, 2001
Time: 1:341:49 PM
*
I 1 T T I
64
Legend
Lon* Configuration
XX/XX Totol Right Turn Vohw/Right Tirn on Rod Votune
XX Opposing Through Volunt
:Xj Conflicting Pedestrians
64
DATA SET: #20
Location: Market Street/18th Street
g/C ratio: 0.50
Date: July 8, 2001
Time: 1:341:49 PM
15/1
lgend
Lon* Configuration
XX/XX b Total Right Turn Voluno/RIght Turn on Rtd Voluiw
XX e Opposing Through Votunc
liX:! B Conflicting Pedestrians
65
DATA SET: #21
Location: Blake Street/19th Street
g/C ratio: 0.50
Date: July 8, 2001
Time: 12:3512:50 PM
J
ngji Â£
7/4 m
n
Legend
Lone Configuration
XK/KK Total Right Tir*n VoUew/Right Ti^n on Red Volune
XX Opposing Through Volune
\Mi Conflicting Pedestrions
66
DATA SET: #22
Location: Market Street/19th Street
g/C ratio: 0.50
Date: July 8, 2001
Time: 12:501:05 PM
No Scale
Legend
Lane Configuration
XX/XX Total Right Turn Volune/Right Turn on Red Volune
XX Opposing Through Volune
Conflicting Pedestrians
67
68
69
DATA SET: #25
Location: Champa Street/15th Street
g/C ratio: 0.33
Date: June 2, 2000
Time: 5:45 6:00 PM
No Scole
Legend
b Lane Configuration
JUWOC b Totol Right Turn Voluftf/Right Turn on fled Volume
XX Opposing Through Volune
Conflicting Pedestrians
70
DATA SET: #26
Location: Stout Street/19th Street
g/C ratio: 0.33
Date: July 1, 2001
Time: 1:001:15 PM
I 19th STREET
71
72
DATA SET: #28
Location: Stout Street/18th Street
g/C ratio: 0.33
Date: July 1, 2001
Time: 12:45 1:00 PM
I n
I 49
15/9
No Scale
Legend
Lane Configuration
XX/XX Total Right Turn Volune/Right Tirn on Red Volune
XX Opposing Through Volume
Conflicting Pedestrians
73
DATA SET: #29
Location: Curtis Street/18th Street
g/C ratio: 0.33
Date: July 1,2001
Time: 12:451:00 PM
I vf it r~
1 22 *
Legend
Lon* Configuration
XX/XX Total Right Turn Volunv/Right Tim on Rad Volun*
XX Opposing Through VoUm*
[SI ConFUcting Ptdtstrlons
74
DATA SET: #30
Location: Champa Street/18th Street
g/C ratio: 0.33
Date: July 1, 2001
Time: 1:001:15 PM
8/4
75
Appendix
B. Regression Results
MODEL 1 SUMMARY OUTPUT
Regression Statistics
R Square 0.306
Adjusted R
Square 0.271
Standard Error 5.619
Observations 30
ANOVA
df SS MS  F Significance F
Regression 1 403.416 403.416 12.778 0.001
Residual 29 915.550 31.571
Total 30 1318.967
Coefficients Standard Error (Star Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercepl 0.000 #N/A #N/A #N/A . #N/A #N/A #N/A #N/A
Total Turn Vol 0.285 0.029 9.805 0.000 0.23 0.34 0.23 0.34
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 10.26 1.26 0.23 1.667 1
2 12.82 3.82 0.69 5.000 1
3 5.98 5.02 0.91 8.333 1
4 12.54 3.46 0.63 11.667 1
5 6.84 2.84 0.51 15.000 1
6 13.39 2.61 0.47 18.333 4
7 7.69 1.69 0.31 21.667 4
8 9.97 7.03 1.27 25.000 4
9 8.83 2.17 0.39 28.333 5
10 16.53 3.47 0.63 31.667 5
11 16.53 4.47 0.81 35.000 6
12 15.10 3.10 0.56 38.333 7
13 17.95 8.95 1.62 41.667 7
14 9.69 4.69 0.85 45.000 9
15 3.99 1.01 0.18 48.333 9
16 2.56 1.56 0.28 51.667 9
17 16.53 9.53 1.72 55.000 9
18 1.99 0.99 0.18 58.333 9
19 6.84 5.84 1.06 61.667 10
20 4.27 3.27 0.59 65.000 11
21 1.99 2.01 0.36 68.333 11
22 4.27 4.73 0.86 71.667 12
23 8.83 5.17 0.94 75.000 12
24 15.67 5.67 1.03 78.333 14
25 10.83 18.17 3.29 81.667 16
26 3.99 8.01 1.45 85.000 16
27 2.28 1.28 0.23 88.333 17
28 4.27 4.73 0.86 91.667 20
29 6.84 0.16 0.03 95.000 21
30 2.28 1.72 0.31 98.333 29
77
MODEL 2 SUMMARY OUTPUT
Regression Statistics
Multiple R 0.592
R Square 0.351
Adjusted R
Square 0.303
Standard Error 5.632
Observations 30
ANOVA
df SS MS F Significance F
Regression 2 462.621 231.310 7.293 0.003
Residual 27 656.346 31.717
Total 29 1316.967
Standard
Coefficients Error tStat Pvalue Lower 95% Upper 95% Lower95.0% Upper95.0%
Intercept 2.968 2.190 1.364 0.184 1.505 7.481 1.505 7.481
Total Turn Vol 0.227 0.059 3.815 0.001 0.105 0.349 0.105 0.349
Conflicting Peds 0.010 0.018 0.524 0.604 0.047 0.028 0.047 0.028
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 9.786 0;786 0.145 1.67 1
2 11.579 2.579 0.475 5.00 1
3 7.405 3.595 0.661 8.33 1
4 12.458 3.542 0.652 11.67 1
5 6.837 2.837 0.522 15.00 1
6 12.709 3.291 0.606 18.33 4
7 6.793 0.793 0.146 21.67 4
8 10.131 6.869 1.264 25.00 4
9 8.967 2.033 0.374 26.33 5
10 15.755 4.245 0.781 31.67 5
11 15.613 5.167 0.955 35.00 6
12 14.450 2.450 0.451 38.33 7
13 16.746 7.746 1.425 41.67 7
14 9.971 4.971 0.915 45.00 9
15 6.143 1.143 0.210 48.33 9
16 4.952 3.952 0.727 51.67 9
17 15.927 8.927 1.643 55.00 9
18 4.375 3.375 0.621 58.33 9
19 8.267 7.267 1.337 61.67 10
20 6.264 5.264 0.969 65.00 11
21 3.917 0.083 0.015 68.33 11
22 5.492 3.508 0.645 71.67 12
23 9.834 4.166 0.767 75.00 12
24 15.151 5.151 0.948 78.33 14
25 11.259 17.741 3.265 81.67 16
26 6.104 5.896 1.085 85.00 16
27 4.639 3.639 0.670 88.33 17
28 6.264 2.736 0.503 91.67 20
29 8.314 1.314 0.242 95.00 21
30 4.697 0.697 0.128 98.33 29
7B
MODEL 3 SUMMARY OUTPUT
Regression Statistics
Multiple R 0.679
R Square 0.461
Adjusted R Square 0.399
Standard Error 5.226
Observations 30
ANOVA
dt SS MS F Significance F
Regression 3 608.421 202.807 7.421 0.001
Residual 26 710.546 27.329
Total 29 1318.967
Coefficients Standard Error (Star Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept . 5.065 4.035 1.255 0.221 13.360 3.230 13.360 3.230
# of opposing thru I 2.976 1.288 2.310 0.029 0.328 5.624 0.328 5.624
Total Turn Vol 0.198 0.057 3.506 0.002 0.0B2 0.314 0.082 0.314
Conflicting Peds 0.023 0.018 1.293 0.207 0.060 0.014 0.060 0.014
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 10.667 1.667 0.337 1.667 1
2 8.874 0.126 0.025 5.000 1
3 10.166 0.832 0.168 8.333 1
4 14.333 1.667 0.337 11.667 1
5 7.734 3.734 0.754 15.000 1
6 13.8B7 2.113 0.427 18.333 4
7 6.572 0.572 0.116 21.667 4
8 11.856 5.144 1.039 25.000 4
9 10.439 0.561 0.113 28.333 5
10 17.408 2.592 0.524 31.667 5
11 17.547 3.453 0.698 35.000 6
12 10.050 1.950 0.394 38.333 7
13 18.052 9.052 1.829 41.667 7
14 11.820 6.820 1.378 45.000 9
15 6.591 1.591 0.321 48.333 9
16 5.461 4.461 0.901 51.667 9
17 11.873 4.873 0.984 55.000 9
18 1.789 0.789 0.159 58.333 9
19 8.226 7.226 1.460 61.667 10
20 9.511 8.511 1.719 65.000 11
21 0.679 3.321 0.671 68.333 11
22 4.662 4.338 0.876 71.667 12
23 12.543 1.457 0.294 75.000 12
24 11.047 1.047 0.212 78.333 14
25 13.537 15.463 3.124 81.667 16
26 6.499 5.501 1.111 85.000 16
27 2.079 1.079 0.218 88.333 17
28 3.559 5.441 1.099 91.667 20
29 8.342 1.342 0.271 95.000 21
30 5.194 1,194 0.241 98.333 29
79
MODEL 4 SUMMARY OUTPUT
Regression Statistics
Multiple R 0.682
R Square 0.465
Adjusted R Square 0.403
Standard Error 5.210
Observations 30
ANOVA
dt SS MS F Significance F
Regression 3 613.113 204.371 7.528 0.001
Residual 26 705.854 27.148
Total 29 1318.967
Coefficients Standard Error tStat Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 16.273 5.995 2.714 0.012 3.950 28.597 3.950 28.597
g/C 30.855 13.105 2.354 0.026 57.792 3.917 57.792 3.917
Total Turn Vol 0.184 0.058 3.178 0.004 0.065 0.303 0.065 0.303
Conflicting Peds 0.013 0.017 0.753 0.458 0.047 0.022 0.047 0.022
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 10.796 1.796 0.364 1.667 1
2 6.978 2.022 0.410 5.000 1
3 9.396 1.604 0.325 8.333 1
4 13.410 2.590 0.525 11.667 1
5 8.283 4.283 0.868 15.000 1
6 13.390 2.610 0.529 18.333 4
7 7.868 1.868 0.379 21.667 4
8 11.373 5.627 1.141 25.000 4
9 10.397 0.603 0.122 28.333 5
10 16.161 3.839 0.778 31.667 5
11 16.237 4.763 0.965 35.000 6
12 15.012 3.012 0.611 38.333 7
13 16.880 7.880 1.597 41.667 7
14 11.381 6.381 1.293 45.000 9
15 3.396 1.604 0.325 46.333 9
16 2.400 1.400 0.284 51.667 9
17 11.237 4.237 0.859 55.000 9
18 1.867 0.867 0.176 58.333 9
19 5.045 4.045 0.820 61.667 10
20 3.441 2.441 0.495 65.000 11
21 1.257 2.743 0.556 68.333 11
22 2.411 6.589 1.335 71.667 12
23 11.553 2.447 0.496 75.000 12
24 15.603 5.803 1.176 78.333 14
25 12.625 16.375 3.319 81.667 16
26 8.591 3.409 0.691 85.000 16
27 7.347 6.347 1.287 88.333 17
28 8.686 0.314 0.064 91.667 20
29 10.354 3.354 0.680 95.000 21
30 7.423 3.423 0.694 98.333 29
BO
MODEL 5 SUMMARY OUTPUT
Regression Statistics
Multiple R 0.680
R Square 0.463
Adjusted R Square 0.401
Standard Error 5.220
Observations 30
ANOVA
df SS MS F Significance F
Regression 3 610.537 203.512 7.469 0.001
Residual 26 708.430 27.247
Total 29 1318.967
Coefficients Standard Error tStat Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 2.844 3.222 0.883 0.386 9.468 3.780 9.468 3.780
Turn Lanes 5.731 2.460 2.330 0.028 0.675 10.787 0.675 10.787
Total Turn Vol 0.199 0.056 3.524 0.002 0.083 0.315 0.083 0.315
Conflicting Peds 0.014 0.017 0.845 0.406 0.049 0.021 0.049 0.021
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 10.852 1.852 0.375 1.667 1
2 6.535 2.465 0.499 5.000 1
3 9.407 1.593 0.322 B.333 1
4 13.732 2.268 0.459 11.667 1
5 8.123 4.123 0.834 15.000 1
6 13.682 2.318 0.469 18.333 4
7 7.629 1.629 0.330 21.667 4
8 11.514 5.486 1.110 25.000 4
9 10.332 0.668 0.135 28.333 5
10 16.700 3.300 0.668 31.667 5
11 16.786 4.214 0.853 35.000 6
12 15.448 3.448 0.698 38.333 7
13 17.478 8.478 1.715 41.667 7
14 11.416 6.416 1.298 45.000 9
15 8.504 3.504 0.709 48.333 9
16 4.560 3.560 0.720 51.667 9
17 11.227 4.227 0.855 55.000 9
18 1.110 0.110 0.022 58.333 9
19 4.545 3.545 0.717 61.667 10
20 2.814 1.814 0.367 65.000 11
21 0.422 3.578 0.724 68.333 11
22 4.518 4.482 0.907 71.667 12
23 11.638 2.362 0.478 75.000 12
24 13.353 3.353 0.678 78.333 14
25 12.784 16.216 3.281 81.667 16
26 5.582 6.418 1.299 85.000 16
27 7.097 6.097 1.234 88.333 17
28 5.680 3.320 0.672 91.667 20
29 10.348 3.348 0.677 95.000 21
30 7.183 3.183 0.644 98.333 29
81
MODEL 6 SUMMARY OUTPUT
Repression Statistics
Multiple R 0.722
R Square 0.521
Adjusted R Square 0.421
Standard Error 5.131
Observations 30
ANOVA
dt SS MS F Significance F
Regression 5 687.171 137.434 5.221 0.002
Residual 24 631.796 26.325
Total 29 1318.967
Coefficients Standard Error rsraf Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 6.157 12.639 0.4B7 0.631 19.929 32.243 19.929 32.243
Turn Lanes 0.407 4.335 0.094 0.926 8.539 9.353 8.539 9.353
Total Turn Vol 0.173 0.058 2.996 0.006 0.054 0.293 0.054 0.293
Conflicting Peds 0.022 0.018 1.256 0.221 0.058 0.014 0.058 0.014
g/C 22.006 21.193 1.030 0.309 65.748 21.732 65.748 21.732
tt of opposing thru lanes 2.178 1.471 1.481 0.152 0.858 5.213 0.858 5.213
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 11.227 2.227 0.477 1.667 1
2 5.959 3.041 0.651 5.000 1
3 10.9B9 0.011 0.002 8.333 1
4 14.600 1.400 0.300 11.667 1
5 8.617 4.617 0.989 15.000 1
6 14.127 1.873 0.401 18.333 4
7 7.458 1.458 0.312 21.667 4
8 12.377 4.623 0.990 25.000 4
9 11.161 0.161 0.035 28.333 5
10 17321 2.679 0.574 31.667 5
11 17.454 3.546 0.760 35.000 6
12 11.701 0.299 0.064 38.333 7
13 17.850 8.850 1.B96 41.667 7
14 12.432 7.432 1.592 45.000 9
15 4.680 0.320 0.069 48.333 9
16 3.477 2.477 0.531 51.667 9
17 9.281 2.281 0.489 55.000 9
18 0.462 0.538 0.115 58.333 9
19 5.675 4.675 1.002 61.667 10
20 6.381 5.381 1.153 65.000 11
21 0.598 4.S98 0.985 68.333 11
22 2.618 6.382 1.367 71.667 12
23 13.171 0.829 0.178 75.000 12
24 12.485 2.485 0.532 78.333 14
25 14.009 14.991 3.212 81.667 16
26 8.129 3.871 0.829 85.000 16
27 4.872 3.872 0.830 8B.333 17
28 5.970 3.030 0.649 91.667 20
29 9.934 2.934 0.629 95.000 21
30 7.182 3.182 0.6B2 98.333 29
62
MODEL 7 SUMMARY OUTPUT
Regression Statistics
Multiple R 0.653
R Square 0.427
Adjusted R
Square 0.384
Standard En 5.292
Observation 30
ANOVA
df SS MS F Significance F
Regression 2 562.726 281.36 10.05 0.0005
Residual 27 756.241 28.01
Total 29 1318.967
Coefficients Standard Error (Stat Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 4.415 4.054 1.089 0.2B6 12.732 3.902 12.732 3.902
TTV 0.191 0.057 3.357 0.002 0.074 0.308 0.074 0.308
Opp Lane 2.428 1.232 1.971 0.059 0.099 4.955 0.099 4.955
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted RTOR Residuals Standard Residuals Percentile RTOR
1 12.2 3.2 0.6 1.7 1
2 9.3 1.7 0.3 5.0 1
3 11.5 2.5 0.5 8.3 1
4 13.7 2.3 0.4 11.7 1
5 9.9 5.9 1.2 15.0 1
6 14.3 1.7 0.3 18.3 4
7 17.3 8.3 1.6 21.7 4
8 10.5 4.5 0.9 25.0 4
9 12.0 5.0 1.0 28.3 5
10 11.2 0.2 0.0 31.7 5
11 11.8 6.8 1.3 35.0 6
12 16.4 3.6 0.7 3B.3 7
13 16.4 4.6 0.9 41.7 7
14 10.6 1.4 0.3 45.0 9
15 5.5 0.5 0.1 48.3 9
16 4.6 3.6 0.7 51.7 9
17 11.5 4.5 0.9 55.0 9
18 1.8 0.8 0.2 58.3 9
19 7.5 6.5 1.3 61.7 10
20 8.2 7.2 1.4 65.0 11
21 1.8 2.2 0.4 6B.3 11
22 5.7 3.3 0.6 71.7 12
23 11.2 2.8 0.5 75.0 12
24 11.0 1.0 0.2 78.3 14
25 12.6 16.4 3.2 81.7 16
26 5.5 6.5 1.3 05.0 16
27 2.0 1.0 0.2 00.3 17
28 3.3 5.7 1.1 91.7 20
29 7.5 0.5 0.1 95.0 21
30 4.4 0.4 0.1 98.3 29
83
MODEL 8 SUMMARY OUTPUT
Repression Statistics
Multiple R 0.978
R Square 0.957
Adjusted R Squ; 0.914
Standard Error 1.012
Observations 9
ANOVA
df SS MS F Significance F
Regression A 91.461 22.865 22.340 0.005
Residual 4 4.094 1.024
Total 8 95.556
Coefficients Standard Error rstat Pvalue Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 1.873 1.155 1.622 0.180 5.078 1.333 5.078 1.333
Opp. Volume 0.064 0.020 3.235 0.032 0.119 0.009 0.119 0.009
Total Turn Vol 0.124 0.024 5.217 0.006 0.058 0.190 0.058 0.190
Conflicting Peds 0.044 0.007 6.286 0.003 0.024 0.063 0.024 0.063
Turn Lanes 6.142 1.160 5.295 0.006 2.921 9.362 2.921 9.362 .
RESIDUAL OUTPUT PROBABILITY OUTPUT
Observation Predicted TOR Residuals Standard Residuals Percentile TOR
1 9.352 0.352 0.493 5.556 1
2 4.547 0.453 0.633 16.667 1
3 2.593 1.593 2.227 27.778 1
4 7.039 0.039 0.054 38.889 1
5 1.124 0.124 0.174 50.000 4
6 0.813 0.187 0.262 61.111 5
7 0.164 0.836 1.169 72222 7
8 4.055 0.055 0.077 83.333 9
9 8.313 0.687 0.961 94.444 9
84
References
AbuLebdeh G., Benekohal R.F., and AlOmari B. (1997) Models for RightTurnon
Red and Their Effects on Intersection Delay. Transportation Research Record
1572. Transportation Research Board. Washington, D.C. pp.131139.
Luh, J. and Lu, Y. (1990) Capacity Computations of RightTurnOnRed Using the
Highway Capacity Manual. Institute of Transportation Engineers Journal, April
1990, pp. 3439.
McTrans (2000) Highway Capacity Software. Gainesville, FL.
Strong Concepts (2000) Signal 2000. Northbrook, IL.
Tarko, Andrzej P. (2001) Predicting Right Turns on Red. Transportation Research
Record 1776. Transportation Research Board. Washington, D.C. pp. 138142.
TRB (1985) Highway Capacity Manual. Transportation Research Board.
Washington, D.C.
TRB (1994) Highway Capacity Manual. Transportation Research Board.
Washington, D.C.
TRB (2000) Highway Capacity Manual. Transportation Research Board.
Washington, D.C.
Trafficware (2001) Synchro 2001. Albany, CA.
Virkler, M.R., and R.M. Ramana (1995) Capacity for Right Turn On Red.
Washington, D.C. Transportation Research Record 1484. Transportation Research
Board, pp. 6672.
85
