1994 Changes to Chapter 9 of the 1985 Highway Capacity Manual
(Signalized Intersections)
Dennis W. Strong, P.E.
The Highway Capacity Manual (HCM)1
is published by the Transportation Research Board (TRB) under the
direction of the Committee on Highway Capacity and Quality of
Service (#A3A10). The latest version of this document was
published in 1985 in looseleaf form to allow for more frequent
updates than have occurred previously. Chapter 9 of the HCM
provides detailed methodologies and procedures for the analysis
of capacity, delay and level of service for signalized
intersections. This chapter is possibly both the most complicated
and the most used chapter of the Manual, which has led to
numerous critical reviews and suggestions for changes of the
Chapter2. Considerable research funding has also been
directed at signalized intersections. All of these considerations
have led to the 1994 publication of a major update to Chapter 9,
along with seven other chapters of the HCM. The subject of this
paper is a discussion of the major changes which have occurred in
this new version of Chapter 9.
A vast number of significant changes have been
made to Chapter 9 of the HCM in this 1994 update. These changes
are a result of the efforts of the Signals Subcommittee of
Committee A3A10 to review the many research reports and other
suggested changes which have been made over the years, and to
assemble these changes into a cohesive update of Chapter 9 which
would advance the state of the art for capacity analysis of
signalized intersections. These changes may be categorized into
three major categories 1) new methodologies and procedures
which constitute a major change to the way an analysis is
conducted, 2) modifications to current methodologies and
procedures which result in minor but important changes to
existing methodologies, and 3) clarifications which provide
clearer instructions and/or extensions to existing methodologies
to remove ambiguities which may have existed in the existing
procedures. Table 1 below lists the major changes to Chapter 9
according to these major categories. Each of these major changes
are described in detail in the paragraphs which follow.
New Methodologies Progression Factor Permitted Left Turn Factor Defacto Left Turn Check ProtectedPermitted Left Turn Planning Method Modifications Clarifications |
Table 1 - Summary of Changes by Category |
NEW METHODOLOGIES AND PROCEDURES
New methodologies and procedures are those changes which constitute a major change to the way an analysis is conducted. Five major changes to these methodologies have been made in the areas of progression factors, left turn analyses and planning methods, as described below.
Progression Factor
A major concern regarding progression factors
(PF) in Chapter 9 is that the PFs in Table 913 are based on
limited field data, and due to the wide range of PF values (0.40
to 1.85), selection of a range of reasonable PFs from the table
often results in a change of the level of service. To address
these concerns, a major research study was commissioned by TRB
which was completed in 19893. The objective of this
NCHRP Project #328C was to evaluate the effects of changes in
the quality of traffic signal progression on stopped delay. The
general approach was to combine the findings of theoretical
modeling with those from field studies and controlled simulation
to obtain a comprehensive evaluation of these effects. A pilot
and four field studies were conducted to collect the data
necessary to calibrate the analytical progressiondelay models.
A significant improvement these models now
provide is a continuous relationship between the variables rather
than the discrete thresholds currently defined, in addition to
their inherently better results due to more significant field
evaluation. A significant change in the application of PFs is
that they are now only applied to the uniform delay term of the
HCM's delay equation. Another significant
change is the general use of the proportion of total arrivals
which arrive on green, used in place of the platoon ratio, to
estimate the PF. A new calibration factor is used for the second
term, as well.
Another major improvement in Table 913 is a
more logical coordination of the effects of progression with the
effects of actuation, as well as the ability to apply PFs to any
lane group, including left turns. The revised Table 913 has
also been closely coordinated with similar changes which have
been made to Chapter 11 (Urban and Suburban Arterials) of the
HCM.
Permitted Left Turn Factor
Another major concern with Chapter 9 was with
the permitted left turn model which is used, one that frequently
produces overly optimistic results. To address these concerns, an
FHWA contract was initiated4 with an objective to
evaluate and improve the HCM model for estimating the impact of
sharedpermissive left turn movements on lane group saturation
flow rate. The study, which was completed in 1989, was based on a
nationwide data collection effort at 25 intersections in four
regional areas and recommended simplified regression models which
differed substantially from the analytical models in the HCM.
After much debate within the Signal Subcommittee and the A3A10
Committee at large, a new "hybrid" model was proposed5
which retained many of the attractive elements of the HCM's
original analytical model while improving the overall results of
the model by using some of the field calibrated regression
results of the original FHWA study. This resultant model was
accepted by the A3A10 committee.
As noted above, these results were limited to
permissive left turns made from lanes shared with through
movements, but the needs of Chapter 9 required a consistent
method for permissive left turns made from exclusive turn lanes.
To address this need, a Left Turn Analysis Task Group of the
Signal Subcommittee was assembled to extend the logic of the
hybrid model to include this situation, as well as other
specialized cases. The result of this substantial effort was an
entirely new permissive left turn methodology which is expected
to deliver more realistic results in general, and in particular,
for some of the more extreme left turn situations. This final
model also has implications which reach further than just the
basic left turn analysis, as described below.
Defacto Left Turn Check
A problem often encountered with the defacto
left turn check for sharedpermissive left turns was that the
defacto check did not indicate an exclusive lane analysis while
the actual left turn analysis did, or vice versa. This was due to
completely independent techniques which were used for the defacto
check and the actual analysis. This situation has been corrected
by a technique employed in the new permissive left turn analysis
described above. During the course of the analysis, an
intermediate calculation is made which determines the proportion
of left turners in the leftmost through lane of the lane group.
If this proportion is 100 percent, then the leftmost lane is
declared a defacto left turn lane and the analysis is restarted
with this new lane group definition. Using this process, the
dilemma of the current manual is avoided and consistent results
are always obtained. This comes at the cost of a few more
computations than before, but it is anticipated that this will be
performed by computer software in most cases, with little or no
impact on the user.
ProtectedPermitted Left Turn
One of the most pervasive problems of Chapter 9
is that of analyzing protectedpermitted left turns. It is this
author's experience that this is the most common difficulty
people have about Chapter 9 analyses and the one that generates
the most discussion about the "correct" way to conduct
an analysis. The reason for this is that it is fundamentally
impossible to perform a "correct" analysis with the
tools presented in the 1985 HCM, compounded by a number of errors
in the example of the HCM which describes how to do the analysis.
This problem has been discussed ad nauseam, with maybe the most
complete discussion presented by Bonneson and McCoy6.
The basic problem is that the delay equation used in Chapter 9
assumes a single green interval during which a constant
saturation flow rate is available. In the case of
protectedpermitted left turns, this is clearly not the case,
having two distinct green periods during which substantially
different saturation flows are available. Much of the discussion
about this problem has centered on how to make the most
reasonable calculation using the 1985 HCM equations. The
modifications made in the 1994 update allow a whole new approach
which recognizes the dual phase nature of the problem.
The Left Turn Analysis Task Group refined the
techniques presented by Bonneson and McCoy using the concepts
described by Hagen7. Two key principles were added,
these being that the analysis should represent the actual timings
provided by the analyst, not an attempt to determine the best
timings, and that the method should provide a distinction between
protectedpermitted and permittedprotected phasings. The
resulting methodology thus provides a completely new, clear and
concise procedure for calculating capacity, V/C and delay for
protectedpermitted and permittedprotected left turn
treatments, all in a manner consistent with the newly adopted
permitted left turn methodology described earlier.
Planning Method
A point of consistent frustration in using
Chapter 9 procedures is with the planning method which is
presented. It has been felt by many that the method is too vague,
and that the implicit levels of service which are determined are
too often in conflict with a more rigorous operations analysis
which may be done subsequently. As a result, the current planning
method is all but unused, with analysts deferring to other
criticallanebased methods such as TRB Circular 212 or locally
generated methods. With these problems in mind, a completely new
planning method has been adopted in Chapter 9 which is closely
patterned after one which has been developed for use in the state
of Florida. This method produces intersection status much like
the old method, but adds a new status called "At
Capacity". It also bases the status on a new variable called
the critical V/C ratio which is an indicator of overall
sufficiency of the intersection geometrics, and is quite similar
to variables used in other methods such as the ICU. This method
is essentially based on the use of default values for many of the
variables of the operations analysis, with the specific intent of
generating results which are more consistent with the actual
operations analysis.
An additional feature of the new method is that
a reasonable and effective signal timing plan can be produced,
including phasing, which can then be used as the basis of a
complete operational analysis which will produce the normal
levelofservice results. Although this estimated timing plan
is by no means an optimization of the intersection's operation,
it is felt that for planning purposes it will be good enough to
provide the basis for generating a delaybased, level of service
result from the planning analysis, a result which has been in
great demand since the release of the 1985 HCM.
MODIFICATIONS
Modifications are those changes to current
methodologies and procedures which result in minor but important
changes to existing methodologies. Two major modifications have
been made in the areas of ideal saturation flow rates and heavy
vehicle equivalency, as described below.
Default Ideal Saturation Flow Rate
Many surveys which have been conducted since
the release of the 1985 HCM have indicated that the ideal
saturation flow rate for a signalized analysis should be higher
than the 1800 vphgpl indicated in the HCM. Although this has
always been a variable which could be changed by a user based on
local conditions, the default ideal saturation flow rate in the
new Chapter 9 has been changed from 1800 to 1900 vphgpl to
reflect these surveys. This is also consistent with similar
changes which have been made in other chapters of the HCM, and is
believed to generate more reasonable results when local data is
not available. Local surveys of both ideal and adjusted
saturation flow rates are still recommended whenever possible.
Heavy Vehicle Equivalency Factor
The basic heavy vehicle equivalency factor used
in the current Chapter 9 is 1.5, meaning that each heavy vehicle
is equivalent to 1.5 passenger car units in terms of saturation
flow rates. Surveys which have been conducted in recent years
have indicated that this value is lower than observed, so the
basic heavy vehicle relationships in chapter 9 have been changed
to reflect a new equivalency factor of 2.0. This is also
consistent with changes made to other chapters of the HCM.
CLARIFICATIONS
Clarifications are those changes which provide
clearer instructions and/or extensions to current methodologies
to remove ambiguities which may have existed in the current
procedures. Nine major areas where such changes have been made
are identified here, although many other changes of a minor
nature have also been made. These notable changes are in lane
utilization factors, right turn on red, improved tabulations and
formulas, critical movement definition, analysis when V/C exceeds
1.0, default values, saturation flow rate surveys, editorial
changes and interpretations/errata, as described below.
Lane Utilization Factors
The definition of and intended use of lane
utilization factors (LUF) has been a vague commodity since the
initial release of the 1985 HCM. Two particular problems were
whether LUFs should be used for lane groups which included any
turning vehicles, and whether they should be used at all for a
typical analysis. While investigating this problem, it was
determined that Case 8 (dual turn lanes) of the left and right
Turn factor tables were actually accounting for lane utilization,
but that the effect was being applied to saturation flow, not
demand volume as with through lanes. As a result, each of these
Case 8 conditions have been removed from the turn factor tables
and now it is recommended that all lane groups with more than one
lane should make use of the LUFs. It is also important to note
that the current version of the Chapter implies that LUFs should
only be used when the conditions of the "worst" lanes
of each lane group are to be determined. The language now clearly
indicates that LUFs should be used under virtually all
circumstances.
In addition, experience has shown that the use
of surveyed LUFs can generate results which are much more
consistent with field observations, so a methodology has been
included in the new Chapter which shows how to survey and
calculate LUFs for use in the analysis.
Right Turn on Red
Treatment of rightturnonred (RTOR)
vehicles has always been an unclear entity in the 1985 HCM, even
after the errata of 1986 was published. Although it is clear that
some treatment of these vehicles is necessary, it was still not
clear what is appropriate. The new Chapter addresses this with a
recommendation which is somewhat more limiting than the current
Chapter, suggesting that RTOR vehicles be removed from the
analysis at the start, but only when the actual number of RTOR
vehicles has been surveyed. One exception has been allowed, that
being when the right turn is from an exclusive turn lane and is
"shadowed" by an exclusive left turn phase on the
crossing street. This recommendation all but eliminates the use
of RTOR in any type of planning analysis where RTOR cannot be
counted.
Improved Tabulations and Formulas
An area of Chapter 9 which has been frustrating
to many analysts is in the table lookup procedures and the
subsequent need for interpolation which is required for any kind
of a precise analysis. Although this problem is often relegated
to computer software, further frustration can set in when one
program uses one table lookup/interpolation strategy and
another uses a different strategy, leading to unexplainable
different results. The problem is exacerbated when extrapolation
is required, where some programs will not allow extrapolation
while others will use differing extrapolation logic. The final
problem in this arena is where tabulated values do not agree with
the equations which might be provided, and where the precision of
the tabulated values is not enough to prevent aggravating
rounding problems.
To address all of these problems, several new
strategies were implemented in all numerical tables of the new
Chapter. First, each table presents the equation which was used
to produce the table, thereby effectively eliminating the need to
interpolate in any case, since each equation is a continuous
function. Second, the limits of many of the tables were better
defined and expanded to encompass all possible input values,
thereby effectively eliminating the need for any extrapolation.
In the few cases where this was not possible, explicit
instructions were provided for extrapolation, the most notable
case being in the lane width adjustment factor. Thirdly, the
precision of all tabulated values was increased to one decimal
place beyond that which would normally be used so that any
rounding errors which are created in the analysis are at a
precision where the error does not create any analytical
problems. Lastly, every attempt has been made to verify the
accuracy of the tabulated values so that they are in exact
agreement with the published equations. In this same arena, a
major simplification of the right turn adjustment table has
attempted to clarify the relationships of the variables and
equations with each of the cases.
Critical Movement Definition
The discussion of how to determine critical
movements and the related lost times for the critical path and
critical V/C calculation has been substantially expanded to cover
the less obvious and more complex multiphase cases. This is
intended to eliminate the great amount of confusion which has
surrounded this area of Chapter 9 over the last nine years.
Particular attention is now paid to how and where movement lost
times should be applied in the analysis, particularly for
protectedpermitted and lead or lag cases. This improvement will
hopefully curtail many of the problems users have had in
understanding the analysis of lost time and critical V/C.
Analysis When V/C Exceeds 1.0
An area of the Chapter which has always been
lacking is for the analysis of conditions under which V/C exceeds
1.0. With the new Chapter, this will still be the case, but a
number of clarifications have been made in the text to better
describe the conditions which cannot be analyzed accurately, as
well as what options the user has. The most significant of these
is with the somewhat arbitrary limitation of 1.2 for V/C. In
reality, the V/C cannot exceed 1/PHF (the inverse of the peak
hour factor), or a condition where the hourly volume exceeds the
hourly capacity will result, a condition which is clearly beyond
the ability of the Chapter to address with any accuracy. Other
related clarifications in the text have been made.
Default Values
The table of default values has been expanded
to include analysis variables which were previously omitted, as
well as new variables which have been added by the new
methodologies described earlier. In some cases, the definition of
the default values has been clarified so as not to cause
confusion created by the earlier table. Separation of the
defaults according to the operational versus planning analyses
has also been provided.
Saturation Flow Rate Surveys
A critical discussion of and proposed
improvements to the saturation flow rate survey techniques
provided in Appendix 4 was written which has now been
incorporated into Chapter 9. These additions provide a clearer
description of what should be surveyed and when certain events in
the survey are defined to have occurred. These changes will
provide a more consistent base procedure to be used in future
saturation flow rate surveys, thus resulting in a more consistent
base of survey data to be used in future editions of the HCM.
Editorial Changes
A great number of editorial changes have been
made throughout the text with the intent of clarifying the
methodologies of the Chapter. In some cases, additional tutorial
material was added to better define the conditions which are
being analyzed, but the greatest portion of these changes are
directed at a more precise definition of the procedures, with the
objective of reduced ambiguity and increased clarity.
Interpretations and Errata
Over the years, a number of interpretations and
errata have been generated by the A3A10 Committee for Chapter 9.
In all cases where these publications do not conflict with other
changes to the methodologies of the Chapter, additional text has
been added to the Chapter to clarify the problems which these
interpretations and errata address.
CONCLUSIONS
This paper has described many important,
significant and useful changes which have been made to Chapter 9
of the Highway Capacity Manual. These changes incorporate
new research and fix identified problems with the existing
materials of Chapter 9, and as such are intended to advance the
state of the art of signalized intersection capacity analysis.
These changes are also designed to remove some of the ambiguities
which have existed in the Chapter, thereby allowing users and
related software to produce more consistent analyses. Software
such as the HCS from McTrans for capacity analysis and
SIGNAL85/TEAPAC from Strong Concepts for optimization using the
new methods is anticipated to be released concurrently with the
updated HCM document. Other packages have also indicated their
intent to update to the new procedures.
It should be noted that not all recognized
needs in the Chapter have been addressed by these changes, with
many of these being under active research or in search of funding
for related research. Most notably, considerations of
oversaturated conditions and the affects of actuations on
intersection performance are significantly lacking and currently
under study. Other areas of need have yet to be addressed
directly, but are under active discussion within the A3A10
Committee and Signals Subcommittee. Furthermore, the new
methodologies introduced as part of these changes have the
distinct possibility of raising new issues which further research
may need to address. As such, the Highway Capacity Manual
will continue to be a living document, with the attempt always to
put the best and most current research in front of the end user
for practical application in the "real world".
Suggestions for such future changes will be readily accepted by
the author, or any other members of the A3A10 Committee.
ACKNOWLEDGMENTS
The author would like to acknowledge the
contributions of the A3A10 Committee, and in particular the
members of the Signals Subcommittee for their hard work and
diligence in the preparation of the materials included in this
major update to Chapter 9. As a completely volunteer organization
with no compensation for their valuable time, the preparation of
this update has been an enormous task which could not have been
completed without the dedication and commitment of these
professionals. We would also like to acknowledge the members of
the subcommittees on Unsignalized Intersections (Chapter 10) and
Urban and Suburban Arterials (Chapter 11) for their efforts to
coordinate the changes between these three highly interrelated
chapters.
REFERENCES
1. "Highway Capacity Manual,"
Transportation Research Board, Special Report 209 (1985).
2. Strong, D.W., "RealWorld Use of the
1985 HCM Analysis of Signalized Intersections,"
Compendium of Technical Papers, Institute of Transportation
Engineers, Washington, D.C. (September, 1989).
3. Fambro, D.B., Chang, E.C.P., and Messer,
C.J., "Effects of the Quality of Traffic Signal Progression
on Delay," National Cooperative Highway Research Program,
Project 328C, Texas Transportation Institute, Texas A&M
University, College Station, Texas (August, 1989).
4. Roess, R.P., Papayannoulis, V.N., Ulerio,
J.M., and Levinson, H.S., "Levels of Service in
SharedPermissive LeftTurn Lane Groups at Signalized
Intersections," Federal Highway Administration, Contract No.
DTFH87C00012, Transportation Training and Research Center,
Polytechnic University, Brooklyn, New York (September, 1989).
5. Prassas, E.S. and Roess, R.P., "The
LeftTurn Adjustment for Permitted Turns from Shared Lane
Groups: Another Look," a paper presented at the midyear
meeting of TRB Committee A3A10, Transportation Training and
Research Center, Polytechnic University, Brooklyn, New York
(July, 1992).
6. Bonneson, J.A. and McCoy, P.T.,
"Operational Analysis of Exclusive LeftTurn Lanes with
Protected/Permitted Phasing," Transportation Research Record
1114, Transportation Research Board, Washington, D.C.
7. Hagen, L.T., "A Comparison of
Intersection Analysis Techniques," University of Florida
(Fall, 1987).
Reprinted with permission,
Institute of Transportation Engineers (ITE), 1994
this page last updated September 22, 1998