Recent and Pending Changes to Chapter 9 of the
Highway Capacity Manual

(Signalized Intersections)

Dennis W. Strong, P.E.
President, Strong Concepts
Chairman, HCM Signals Subcommittee

 

Abstract:

The Highway Capacity Manual (HCM) is published by the Transportation Research Board as a means of standardizing the techniques used to evaluate the quality of service provided by various transportation facilities. 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 the most complicated and probably the most used chapter of the Manual, and considerable research funding has been directed at signalized intersections. These considerations have led to the recent 1997 publication of a major update to Chapter 9, with another update pending for 2000. The subject of this paper is a summary of the major changes which have occurred in these new versions of Chapter 9, and a reflection on how these changes will manifest themselves in our daily work. A technical paper focused on the details of the changes was presented by the author at the 79th Annual Meeting of the Transportation Research Board this past January.

1997 Changes to the Highway Capacity Manual

A number of significant changes have been made to Chapter 9 of the HCM in the 1997 update. These changes are a result of the efforts of the HCM Signals Subcommittee to review research reports and other suggested changes made since the 1994 HCM was published, 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 five major areas:

Each of these major areas of change and their impact are described in the following paragraphs.

New Use of Control Delay

Prior versions of the Highway Capacity Manual (1985 and 1994) used 'stopped delay' as the measure of effectiveness (MOE) for the performance of signalized intersections. This was due to a perception that stopped delay, the amount of time a vehicle was physically stopped on the pavement, was an easier performance measure for a driver to perceive and for traffic personnel to survey. The difficulty which resulted was that virtually all other related methodologies and software used 'total delay', which made comparison of the HCM results to other models difficult. Even other chapters of the HCM such as Unsignalized Intersections (Chapter 10) and Arterial Streets (Chapter 11) used total delay as the primary MOE. In order to be more consistent with all of these other methodologies, it was decided that the 1997 HCM should also use total delay as its MOE for signalized intersections, so the necessary adjustments were made. Effectively, the divisor of 1.3 which had previously been used to convert a total delay model to represent stopped delay was simply removed from the delay formulas and the delay thresholds for level of service were increased by this same 1.3 factor. New survey techniques were also developed to survey the new total delay quantity, and it turns out these techniques are no more difficult than those used previously. The new delay value is also being called 'control delay' to reflect that it is the delay which results from the traffic control device only, and that other delay values such as geometric delay or incident delay will contribute to the real 'total delay'.

This change will result in the reporting of 30 percent higher delay values, with the appearance that the level of service delay thresholds are more lenient than before by allowing these higher delays to result in lower levels of service than before. The reason is, of course, because we're reporting a different delay value than before and the thresholds have compensated for this. For example, with all other concerns remaining equal, a previous calculation of 25.0 seconds of stopped delay will now result in a calculation of 32.5 seconds of control delay (25.0 x 1.3 = 32.5). This is the reason the level of service C threshold has been raised from 25 to 35 seconds (rounded "up" from 32.5). In the short term we'll struggle to get used to these new, higher values and what appears to be a higher tolerance for delay. In reality, this change is more of a cosmetic change that by itself doesn't really have a significant impact of level of service results.

New Model for Oversaturated Delay

A major limitation of the HCM signalized method has been its inability to quantify the delay which resulted from an oversaturated situation. This limitation caused significant problems when evaluating problem intersections, especially when attempting to compare before and after calculations. The problem was in the original formulation of the delay equation used in the 1985 HCM where the manual stated that conditions with volumes that exceeded capacities by more than 20 percent could not be calculated using the delay formula. Since that time the Federal Highway Administration has commissioned a study which determined, in part, that a minor change could be made to the delay equation which would allow oversaturated conditions to be evaluated. With this in mind, the HCM Committee decided that this change should be implemented, along with several related changes. These changes now allow delays to be calculated without any limits on the volume-to-capacity ratio, as well as for conditions where oversaturation exists for more than 15 minutes.

The impact of this change is that users of the HCM will now be able to better and more completely evaluate problem conditions which involve oversaturation, as well as to quantify how much these conditions can be improved. These new procedures, which also now allow analysis for conditions which have a queue at the beginning of the study period, will permit more complete and comprehensive multi-period analyses which span across an entire peak period of time-varying demands, resulting in more accurate and useful peak period analyses for evaluating congestion management alternatives.

Improved Definition and Use for Lost Time

The 1985 and 1994 HCM's were quite casual in their definition and recommendations for use of the lost time quantity needed for calculating signalized intersection delay. The value was properly defined, but since little advice was provided for determining values to be used, the published default value of 3 seconds was frequently used by users of the HCM, even when changes were made to intersection analyses which would affect the lost time. To correct this problem, the inputs required of the user have been changed so that the lost time needed is calculated from these inputs rather than being input directly. These new inputs, startup lost time and extension of effective green, are formulated so that the ending lost time can be calculated from the input value of yellow and all-red time, and is therefore determined by the length of the clearance period. Thus, when a clearance period is changed, the total lost time will change by an equal amount unless the user specifically prevents this from happening. Consequently, when default values of the inputs are used, the lost time will be dependent on the clearance value, a feature which did not exist in previous versions.

The impact of this subtle change to the procedures can be quite significant, and may be quite apparent in capacity analyses performed with both the old and new methods, especially where default values are used. For example, when default values are used in the new method, the total lost time which results for each signal phase will be equal to the total yellow and all-red clearance time for each phase. Since clearance times are frequently found to be 5 to 6 seconds, lost time values in this range for the new analysis will be significantly higher than the default value of 3 seconds found in the old analysis. This can result in a reduction of useable greentime throughout a cycle of as much as 10 or more seconds for multi-phase signals (in comparison to the old analysis) which can be quite significant for medium to short cycles. The net effect can easily be a drop of one or even two levels of service between the two analyses if all other things are equal. This can be quite an eye-opener, all due to a subtle and apparently minor change to the way lost time is determined.

New Treatment of Actuated/Coordinated Movements

The HCM has long-needed an improved way of assessing the impact that actuated movements have on a signal's operation. Previously, the model simply stated that the impact of an actuated movement was a reduction of the primary delay by 15 percent, regardless of the quality or specifics of the actuation provided. Recently a National Cooperative Highway Research Project study was completed which aimed at improving this HCM model. Essentially the recommended changes from this study came in two parts: a change to the delay equation which made it sensitive to the extension value of the actuated movement, and a better way to estimate the average greentimes for an actuated signal which would be used in the capacity analysis. These changes were implemented directly in the 1997 update. This also simplifies the adjustments which are made for coordinated signals, and a related adjustment was added to account for the affect of congested upstream conditions. The result is that our capacity analyses can now reflect to a greater degree how the values of various actuated control parameters such as initial, extension, maximum, detector size, detector setback, etc. affect signalized delay.

Miscellaneous Changes

Several miscellaneous changes were also made which warrant brief mention due to their significance. One is that lane utilization factors are now used to reduce saturation flow instead of fictitiously increasing demand. With this change volume-to-capacity ratios will remain nominally unchanged with a more realistic way to account for unbalanced lane use and no need to create fictitious vehicles. The new method will also result in saturation flow rates which are more appropriate for use in other traffic models such as TRANSYT, PASSER and CORSIM, and will create greater consistency between analyses done with the HCM procedures and these other models

Another change affects several of the factors involved in the permitted left turn model. In this regard, the left-turn equivalency model has been improved, simplified and adjusted to reflect the changes in lane utilization procedures, and several of the other sub-models have been improved with better formulas and more logical limits. The net effect is that the permitted left turn model should deliver more reliable results over a wider range of conditions, thus improving our overall accuracy when using the HCM capacity method under these conditions.

Other minor change included the further clarification of an often-misunderstood reality that an HCM capacity analysis is normally performed for a 15-minute peak. Thus, the use of peak hour factors is merely to estimate these peak 15-minute flow rates when the 15-minute flow rates are not otherwise available. Consequently, since 15-minute counts are typically performed, the flow rates for the intersection's peak 15-minute period should be determined and peak hour factors should not be used (set equal to 1). The use of the CBD adjustment factor was also clarified, and minor errata from the 1994 publication were incorporated.

2000 Changes to the Highway Capacity Manual

A number of important additional changes have been made to Chapter 9 of the 1997 HCM which will appear in a pending 2000 update. These changes include:

Each of these major areas of change and their impacts are described in detail below.

Re-organized Chapters and Editorial Changes

The Highway Capacity Manual 2000 (HCM2000) represents a major overhaul of the three prior versions of the manual released in 1985, 1994 and 1997. The entire HCM has been completely re-organized with over 30 chapters, and will come in both a printed manual and as a multi-media CD-ROM. Many completely new chapters and procedures have been developed for the HCM2000, but the signalized intersection materials are fundamentally the same as in the 1997 version, with the exception of the items discussed below. Even so, the signalized materials have been substantially enhanced with new figures and discussions, and the new layout of the HCM makes the materials much easier to read and understand. Many of the complex procedures of the signalized method have also been relocated to appendices, making the basic procedures easier to follow.

In the new organization of the HCM 2000 the signalized intersection materials have been divided into two chapters. Chapter 10 contains general concepts for all interrupted flow facilities, including signals, stop signs, roundabouts and arterials. This chapter provides discussions of general concepts which are common to more than one interrupted flow facility type, as well as general concepts specific to each individual facility. Chapter 10 also contains the materials which relate to planning uses of the signalized method. Chapter 16 is the chapter which now contains the primary methodology for signalized intersection analysis.

New Model for Queue Lengths

The Highway Capacity Manual has never had a formal queue model which users could use to estimate queue backups for both signalized intersection evaluation and design. This has been a long-standing need of the HCM, and with its continued omission from the manual users have used a wide variety of queue models which deliver widely varying results due to their inconsistent formulations. To address this critical need, the HCM2000 development project conducted a literature review to determine the specific needs users had for a comprehensive queue model, then set out to develop such a model that met these needs. The publication of the HCM2000 will see the fruits of that effort in a comprehensive queue model which includes the effects of specific signalized conditions including greentime, cycle length, phasing, saturation flow rate, capacity, coordination, actuation, oversaturation and initial queue. This model estimates the most important part of the queue, the extent of the back of the queue measured from the stop bar, and will also estimate various degrees of confidence in the form of percentile values.

The existence of a formal queue model in the HCM will provide for the first time a national standard for estimating queue lengths, and as a result will generate some consistency across the country on how this estimation is done. Hopefully, this will also lead to a similar consistency in the queue models used in various software packages, which presently run the gamut of possible queue models. In addition, for the first time we will have a comprehensive queue model that is sensitive to all the signalized conditions we need to consider when modeling queues so we can begin making more informed decisions and designs in this regard.

New Adjustments for Pedestrians and Bicycles

A Federal Highway Administration research project was recently completed which detailed the effect that pedestrians and bicycles have on turning traffic at signalized intersections. The results of this research have been implemented in the HCM2000 in a manner that now allows us to quantify in a much more detailed and precise way how these alternate modes will affect signal operation and performance. These effects, which were previously included in the turning traffic adjustment factors, have now been separated out into their own satflow adjustment factors so they can be modeled separately and the effects of various designs can be observed directly. This is an important step forward for facilities where pedestrians and/or bicycles are a significant part of the total traffic mix.

Protected-Permitted Left Turn Model Changes

Yet another minor improvement has been made to the permitted left turn model in a continuing effort to make this complex model as realistic as possible. In the HCM2000 an older model for treating protected-permitted left turns which make their movement from a shared through lane (called Case 6 turns) has been updated so it is consistent with how both the protected and permitted movements are made on their own from a shared lane, as well as consistent with how a protected-permitted turn is handled from an exclusive turn lane. New terminology for this condition has also been established so it can be concisely described without a mouthful of words: the complete old terminology was "protected-plus-permitted or permitted-plus-protected left turn" where the new terminology is simply "compound left turn".

Summary

The changes made to the signalized intersections methods of the Highway Capacity Manual in 1997 and 2000 constitute major improvements in our ability to model signalized intersection conditions more accurately. As a result, we have been empowered to make better assessments of the impacts of various potential or projected changes to our signalized conditions, as well as to create better designs to accommodate these conditions in the future.

this page last updated March 11, 2004