USF Libraries
USF Digital Collections

A fresh look at roadway level-of-service issues


Material Information

A fresh look at roadway level-of-service issues
Physical Description:
1 online resource (various pagings) : ill. ;
Ewing, Reid H
University of South Florida -- Center for Urban Transportation Research
Center for Urban Transportation Research, University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:


Subjects / Keywords:
Highway capacity -- Mathematical models -- Florida   ( lcsh )
Transportation demand management -- Florida   ( lcsh )
Urban transportation -- Planning -- Florida   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Statement of Responsibility:
by Reid Ewing.
General Note:
Title from cover of e-book (viewed Aug. 3, 2011).

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029308394
oclc - 744665787
usfldc doi - C01-00002
usfldc handle - c1.2
System ID:

This item is only available as the following downloads:

Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200313Ia 4500
controlfield tag 001 029308394
005 20110803155848.0
006 m d
007 cr bn|||||||||
008 110803s199u flua ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a C01-00002
1 100
Ewing, Reid H.
0 2 245
A fresh look at roadway level-of-service issues
h [electronic resource] /
by Reid Ewing.
[Tampa, Fla.] :
b Center for Urban Transportation Research, University of South Florida,
1 online resource (various pagings) :
Title from cover of e-book (viewed Aug. 3, 2011).
Highway capacity
z Florida
x Mathematical models.
Transportation demand management
Urban transportation
University of South Florida.
Center for Urban Transportation Research.
t Center for Urban Transportation Research Publications [USF].
4 856


A Fresh Look at Roadway Level-of-Service Issues by Reid Ewing


A FRESH LOOK AT ROADWAY LEVEL-OF-SERVICE ISSUES Reid Ewing Center for Urban Transportation Research University of South Flo rida ABSTRACI' The tendency in growth management is to focus on roadway level-of-service sta nd ards. However, the methods u sed to determine roadway l evel s of service may affect conclusions about road adequacy as much as do the st andards t o which they are compared. The specific technique used to analyze roadway levels of service can make at least a two-letter grade difference in the outcome Likewise, the choice of analysis period or peak hour can make a difference of two or more letter-grades While harde r to quantify, the effect of averaging/aggregating leve ls of service across facilities could be of comparable magnitude. Thus, even adopting th e same level-of-service. sta nd ards le vel -of-se rvice determinations for, say, the City of Miami and Jefferson County, Florida have entirely different implications for motorists. Jefferson County has opted for a "by the book" approach, comparing the 30th highest hourly traffic volumes on individual r oads to the maximum volumes at different levels of service bas ed on Hif:hway C-apacity Manual methodology. In contrast, Miami has adopted an innovative but unconventional approach, comparing perso n -trip volumes for the two highest hours on the average weekday to the practical capacities o f multimodal transportation corridors. Among the novel approac h es to roadway level-of-service determination reviewed in this article, three seem particularly promising: (1) use of simple regression models to estimate average travel speeds, and from them arteria l levels of service; (2) development of level-of-service meas ures and standards for travel corridors and traffic districts; and (3) use of lOOth highest hourly traffic volumes as the basis for roadway level-of -serv i ce determinations in urban areas


I. INTRODUCTION Roadway levels of service play a central role in Florida's effortS to manage growth. The State's 1985 Growth Management Act embraced a "pay as you grow" philosophy, commonly known as concurrency. Adequate infrastructure must be available concurrent with the impacts of development. Adequacy is defined by level-of-service standards, adopted by local governments as part of their comprehensive plans. No development order or permit may be issued if levels of service will be degraded below the adopted standards. As one observer noted: "Five of the infrastructure elements have posed few problems for local governments. But the sixth category, roads, is proving to be a nightmare."' In 1987, the State Comprehensive Plan Committee projected a $53 billion infrastructure shortfall by the year 2000 --unless growth is slowed or infrastructure investment increased. Of that amount, about half was transportation-related. Roads are the infrastructure element most likely to trigger public dissatisfaction growth moratoria, and legal challenges under the Growth Management Act. Thus, it is crucial that roadway level-of-service determinations be accurate and results be interpreted meaningfully. This paper explores alternative approaches to ensure they will be. Scope of Inquiry Florida jurisdictions take a variety of approaches to roadway level-of-service 1 John Koenig, "Down 10 lhe Wire in Florida; P Mning (October 1990), p. 6. 2


determination (see Table 1). Most jurisdictions go ''by the boo k." They analyze 30th highes t hourly volumes roadway-by -roadway using methodology from the 1985 Highway Capacity Manual. A few jurisdictions have opted for innovative but unconventional alternatives to the standard approach. While it is tempting to reject these alternatives as "not professionally accepted," it must be remembered the was written for applications other than areawide growth management. This relatively new area of application requires fr esh thinking. Accordingly, three old methodologica l issues are addressed anew in this article: (1) W1Jat methods should be used to assess roadway levels of seNice? (2) Should levels of service be averaged or aggregated across road facilities? (3) For what time period or peak hour should levels of service be analyzed? II. METHODS OF ANALYSIS Methods of estimating roadway levels of service may be arrayed in terms of data and analytical requirements, and corresponding precision of estimates. It is sometimes assumed that the simplest methods are the least precise, the most complex methods the most precise (as in Figure 1). This assumption is l argely untested. From least to most complex, methods commonly used in Florida are: Comparison of traffic volumes to maximum volume tables for "generalized" traffic, roadway, and signal conditions. The generalized tables were developed by the 3


Florida Department of Transportat ion (FOOT) based on methodology in the 1985 Hi&hway Capacity Manual (HCM). Levels of service are determined by comparing traffic volumes to the maximum volumes at d i fferent levels of service for roads of a given type. Comparison of traffic volumes to maximum volume tables for typical traffic, roadway and signal co n dit i ons in a given locale. The tables are generated with FOOT computer programs based on 1985 HCM methodo l ogy (LOS and ARTTAB). Again, l evels of service are determined by compari n g traffic volumes to maximum volumes at different levels of service Analysis of levels of service for individual roadways using computer programs based on 1985 HCM methodology (ART-ALL, ART-PLAN and HCS Artedal Analysis) ART-ALL and ART-PLAN are simplified vers i ons of HCS, developed by FOOT. Measureme n t of actual levels of service in the field using travel time and delay studies. Assessmen t o f Alternative Methods To the author's knowledge, there are no pub l ished studies comparing actual roadway levels of service to est i mates obtained by di f fe r ent methods. Reports a n d articl e s comparing traffic models and methods of analysis have stopped short of field testing .2 To assess common met h ods of ana l ysis, traffic and speed data fo r three arterials were 'See, for example, Ahmad Sadegb et al., "A Comparison or Arterial and N e twork Software Programs, ITE Journ a l (August 1987), pp. 3 5; and Dane lsmatt, "A Comparison or the 1985 Highway Capacity Manual and the Signal Operation s Analysis Package 84," Transportation Record 11)91 ( 1987), pp. 109. 4


acquired from consulting firms. Two of the arterials, Kirkman Road and Turkey Lake Road, are in Orlando.3 The former has high traffic volumes and low signal density, the Iauer relatively low traffic volumes and higher signal density. Intersections were first analyzed with HCS, and results were fed into the HCS arterial analysis program to obtain estimates of av erage travel speed for each arterial.' Assumptions from HCS runs were then carried over to ART-PLAN and ART-TAB runs.6 This meant that the three methods of estimating levels of service could be compared to travel time runs with some assurance that all were measuring the same conditions. Estimated and actual average travel speeds are compared in Figures 2 and 3. Given two arterials, two peak periods, and two directions, eight comparisons can be made. It 1 Data w ere supplied by Olatting Lopez Kercher Anglin, Inc., an Orlando-based planning firm. For each arterial, traffic counts and travel time runs were done during the same peak hours on I he same weekdays. Thus, by dcsign, actual travel speeds (derived from travel time runs) and estimated travel speeds (dependent on traffic counts) relate t o the same time pe riods. s In a series of intersection analyses, liberal assumptions were made about; the saturation flow rates at intersection s on these arterials (1,850 vehicles pc:r hour after adjustments), the amount of green time devoted to arterial-through movements (the maximum possible, given the timing plans of these semi actuated traffic signals) arrival types of vehicle platoons (the best possible progression, given signal s paciqg. and signal timing offsets), and the peak-hour factor (a value of 1.0 was assumed, as if flow rates were absolu(ely constant during the peak hour). A peak hour factor of 1.0 was assumed to achieve a measure of consistency between HCS estimates and travel time runs. If actual pe. ak-hour factors had been used instead. HCS eslimates would have applied to the peak 15minutc period of the peak hour, while the travel time results were averaged over t he entire p ea k hour. ARTPLAN and ART TAB represent the latest generation of su ch programs and thus are more fittingly evaluated than are their predecessors, ART ALL and LOS. 5


appears that actual travel speeds are significantly higher than estimated speeds in nearly all cases. They are 5 to 10 mph higher in most cases than HCS-derived travel speeds. The other arterial analyzed was Broward Boulevard in Fort Lauderdale/Plantation. 7 Compared to Kirkman Road, Broward Boulevard has higher traffic volumes on side streets and thus can claim a smaller portion of the signal cycle to accommodate its heavy traffic volumes.8 Estimated travel speeds on Broward Boulevard are a fraction of actual speeds (see Figure 4).9 If results for Kir k man andThrkey Lake Roads suggest that methods of analysis underestimate travel speeds, res ults for Broward Boulevard indicate that methods break down entirely when demands are too heavy relative to intersection capacity. Why Estimates Differ from Actual Travel Speeds To hel p explain why actual trave[ speeds are higher tha n estimates results for Kirkman and Turkey Lake Roads were analy z ed by roadway segment and by component of total travel time the components being delay at intersections and running time between 1 Data were supplied by John Zeegcr of Barlon-Aschman Associates, Inc ... Fort Lauderdale-. Average travel speeds and peak hour volume.< for Broward Boulevard were gathered on comparable weekdays of the same month. Portions of I he cycle devoted 10 the through movements on Broward (so-called green ratios) were observed at the same time traffic counts were taken. T h ey were subsequently confirmed from signal timing plans. Thus, while green r atios for Broward Boulevard appear very low, there is no reason to doubt the validity or the values supplied b y the consulting firm. 9 The section of Broward Boulevard analyzed is between N.\V. 18th Avenue and State Road 7. HCS runs could not be done for Broward Boulevard because information about cross-street traflic and signal timing i s not availab le. 6


intersec t ions.10 Typ i cal results ( f or northbo u nd AM movements ) a r e pre s ented in Tables 2 and 3. S t op p e d d e l ays i n the trave l time runs are mostly s h orter t han estimated with HCS and ART-PLAN Differences are exaggerated for intersections with longer delays. R unning speeds in trav e l time runs are significantly h igher than e stimated with eit her p r og r am, and as such, account for most of t he difference between actual and estimated a v e r age travel sp e eds F ollow in g convent i on, the posted speed limit was taken as the free flow speed i n program runs Ye t roads a r e often des i gned for safe speeds in excess of the posted speed limits, and as casual observati o n sugg e sts, drivers have a tenden cy to driv e at desig n speeds o n lo ng, unin t errup t ed segme n ts wit h m oderate t r affic volumes.11 There is another reason why running speeds in travel time runs are higher than estimated with HCS and ART-PLA N The programs assume some stopped delay at all intersections a n d hence so m e acc e lera t io n and dece l erat i on which depress average running speed s i gnifica ntly o n s hort segments.12 However, no delay was act u a ll y e x perienced on most run s at intersections with high green rat ios and good p r ogression. Hence the free flow 1 From traveltime runs. the s topped delay at intersections and the tota l running time between intersections are known. From HCS runs, estimates of stopped de lay, to tal intersection approach delay, and running t i m e between intersections are available. Stopped de lay f r om travel time runs can be compared directly to CSiimate$ from HCS runs. However, t o compare total running time and hence running speed be t ween intersec t ions. an HCS estimate of r u nning time must bo innated by the dif f erence betwun total approac h delay and stopped delay at the downstream intersection. This adjustment captures the extra running tim e with dece ]erat i o n of vehicles approaching the downstream inte rsec:tion. 11 This is n o t to sanction the practice of speeding b u t to acknowled ge that it occurs. The problem lie s with the level-of-service measu re chosen for arterials in the 1985 HCM. Many alternative me a sures were considered d u ring the up d ate of the HCM. Perhaps a measure s u ch as average delay would have been more s u ita b le th an average travel speed u Delay associated with acceleration is incorporated i nto r unning time estimates for segments of less than one mile (see Tabl e 11 in the 1985 HCM). Delay associated with deceleration is lneorporatcd into inte.rsection approach delay cstim3.te s. 7


speed was maintained over the entire length of segments or even sections. Application of Travel Time Studies Only three of 11 local transportation planners interviewed for this study said their jurisdictions conduct travel time studies as a method of determining arterial levels of service. One reason is the relatively high cost of such studies; this factor may be rendered moot eventually by advances in automatic vehicle locat ion (A VL) technologyY The more important reason is the percept ion t hat travel time study results apply only to the specific time period when travel time runs are done -t hat they cannot be used to predict future levels of service, as req ui red in growth management. This perception is incorrect. T ravel tim e study results can be used to calibrate HCS, ART-PLAN, and other programs. Default values assumed by t he se programs may not be applicable to a particular loca le.14 Programs can be run with progressively higher saturation flow rates or free flow speeds until estimated intersection delays, running speeds, and overall travel speeds better app roximate travel time study resu lts. The better-calibrated programs can then be used to f orecast future levels of service. Alternatively travel time study results can be correlated directly with traf fic volumes l l The City of Miami may eventually outfit its fleet of vehicles with transmitters that supply continuously updated travel speed data. 1 John D. Zegeer. "Field Validation of Intersection Capacity Factors; Transnonation Research Record 1091 (1987), pp 67-77. 8


and other variables in statistically derived models. To i llus trate this approach, average peakhour travel speeds a nd traffic volumes were acquired for 17 two-lane roadways in Seminole County, Florida. With a.m. and p.m. peak hours, and northbound and southbound directions, speed and volume data were available for a total of 68 movements. Average peak-hour travel speed was regressed on peak-hour traffic volume and two other variables, number of signalized intersections per mile and free-flow speed. Both linear and nonlinear forms of the regression equation were tested. The best fit to the data was obtained with a linear equation in two independent variables peak-hour traffic volume and signalized intersections per mile (see Figure 5).15 The explanatory power of the model estimated for Seminole County is probably inadequate for use in forecasting future. travel speeds and levels of service. The standard error of the estimate, 5.3 mph, could result in a oneor even two-letter grade difference between estimated and actual levels of service. Nonetheless, with 55% of the variation in average travel speed explained by only two independent variables, one has to believe that a good predictive model could b e developed with a richer data base (including such independent variables as the green ratio, arrival type, and % turns from exclusive lanes). The regression model' s simplicity may be a virtue. The complicated models and multitude of parameters used in the 1985 HCM methodology only give the appearance of precision. I n light of results for Kirkman Road, Turkey Lake Road, and Broward Boulevard, added complexity may not translate into added precision. u The speed-volume r elationship ls known to become nonlinear as road capacity is approached. Apparently, Seminole Counry roads operare in a now range rhar is adequately approximarcd by a linear equation. 9


Ill. AREAWIDE LEVELS OF SERVICE The Florida Engineering Society Journal (September 1990) featured a debate over the merits of averaging roadway level of service within a corridor, district, or entire urban area. One author contended that averaging could result in a "glossing over of transportation problems."16 Another countered that requiring each roadway link to operate at a minimum acceptable level of service causes "short-term incremental improvements rather than long term comprehensive improvements."" Both authors are right. The challenge is to devise level-of-service measures and standards tha t encourage a long-term comprehensive approach to transportation improvement programming while still addressing localized traffic problems. Current Practice It is routine in traffic impact studies to estimate levels of service for: a lane group at an intersection, an entire inter sect i on, a roadway seg ment from intersec t ion to interse ction, and a section of roadway with multiple intersections along its length. 16 Richard A. "Concurrency Management for Transportation," Florida Ensineeriog Society Journal (September 1990), p. 20. ,, TimothyT. Jackson, "Transporlation Concurrency: How Can h Be Achie v ed? Florida Engineering Socierv Journal (September 1990), p. 24. 10


As we move up the hierarchy from intersections to entire roadway sections, we are averaging or aggregating levels of service. The procedure for doing so is straightforward, at least for urban and suburban arterials. For signalized int ersections, levels of service are measured in terms of average stopped delay. Add to this tbe delay approaching intersections and the running time between intersections, and we obtain total traveltime on a roadway segment. Divide the length of the segme nt by total travel time, and we arrive at average travel speed, which determines the level of service. Do the same for several segments in a series, and we obtain an estimate of level of s ervice for a roadway section. No difficult conceptual issues arise, and no one questions the basic logic of averaging or aggregating levels of service in this context, even though the group o f motorists experiencing the "average" travel speed is constantly changing over the length of the arterial. However, we find ourselves in uncharted waters when we begin to combine levels of service across facilities. There is no s tandar d, professionally accepted method of averaging or aggregating levels of service within: a travel corridor, a traffic district, or an entire road network. Concepts Underlying Aggregation Two distinct concepts may b e used to justify and guide the aggregation of roadway levels of service. The first is the concept of typical trips. Over the course of a day, a 11


person may travel on scores of roadway links and dozens of different roads. Even a single peak-hour trip may involve travel on a myriad of facilities. Presumably, a traveler's perception of roadway conditions is based on an entire trip or possibly even an entire day's worth of travel, not on the delay at one intersection or congestion on one r.;>adway segment. Therefore, roadway levels of service might reasonably be aggregated to reflect common travel patterns and trip lengths. Aggregation may also be justified by the concept of alternate routes. Where a well developed road network exists, an individual may have many routes available for a given trip. Ordinarily, the routes will offer different levels of service since they are made up of segments with varying travel demands upon them from other trip makers. If any route provides an acceptable level of service, government may have met its responsibility to the individual trip maker. Hence, roadway levels of service might reasonably be aggregated across alternate routes within travel corridors. Methods of Aggregation How can levels of service on individual facilities be combined into one areawide level of service? While there is no standard approac h three possibilities suggest themselves. All three have precedents in F l orida's loca l comprehensive plans. The first approach is to sum traffic volumes and capacities for roads in a given area (where capacities are equal to maximum volumes at adopted levels of service). If the sum of traffic volu mes is less than the sum of capacities, the area might be deemed to meet 12


level-of-service standards. Lee County, Florida sums traffic volumes and roadway capacities within traffic districts, and uses any net capacity to justify degradation of already "backlogged" roads (see Figure 6).18 A second approach is to average levels of service across facilities of a given type in an area While averaging in this context is novel, averaging travel speeds on arterials has been an accepted practice since the 1985 HCM was released. It is not difficult conceprually or methodologically to go from averaging speeds on arterials to averaging speeds across arteria ls. The Breva r d County Comprehensive Plan provides for a veraging of levels of service on a main arterial and parallel interconnected collectors. A third approach to aggregation is to adopt a performance summary for r oads in an area, which specifies the percentage of roads at or above given levels of service. Unlike the preceding approach, which applies a performance standard to an "average" roadway, or the conventional approach, which applies a performance standard to each road i ndividually, this third app r oach applies a standard to the performance summary. An example of this approach is found in the Orlando Comprehensive Plan (see Table 4) All three approaches -summing volumes and capacit i es, averaging levels of service, and adopting performance summaries -allow local governments t o finance the most costeffectiv e system im provements rather than iso l ated ro a dway i mprovements dictated by minimum operating standards. How to choose among them? One method -the adoption of performance "FOOT defines backlogged roads as roads on the State Highway System operating below level-ofservice standards and not programmed for improvement in the first three years of FOOT's adopted work program or in the capital improvement element of any loc-al government. 13


summaries -conforms to standard engineering practice, while the other two extrapolate from such practice. There are no professionally accepted level-of-service measures for road networ ks only for individual roads. By continuing to analyze roads individually, performance summaries avoid methodological leaps of faith. Even so, the averaging method may be preferred for growth management purposes. Travel speeds fall precipitously as traffic volumes approach capacities. With areawide averaging, local governments, concerned about maintaining average travel speed, will have considerable incentive to fix traffic hot spots. Less incentive is provided by the other methods of aggregation. The fac t that areawide averaging is not standard engineering practice may have little practical significance. Professionally accepted practices could change with a future update of the Highway_ Capacity Manual, as leve l-of-service standards are increasingly applied to growth management. Even if level-of-service standards remai n t ied to individual f acilities, areaw i de averaging will gain all of the l egitimacy required for growth management in Florida if it is accepted by Florida's state planning agency Weighting Factors Whichever method is chosen, ro adways must be assigned weights that reflect their contributions to overa ll levels of service. Lee County weights traffic volumes and capacities of roadway segments by their respective lengths (i.e., by centerline miles). Brevard County also uses s e gment lengths as a weighting factor. Pasco County w e ights its performance 14


summary by the number of vehicle-miles traveled on different roads ; Orlando uses the number of lanemiles ; and Tampa uses centerline miles in one performance summary and vehicle-miles traveled in another. Use of vehicle miles accounts for the volume of traffic exposed to different traffic conditions. Since it is the "average" experience of travelers we wish to capture in an overall level-of-service measure, not the average condition of roadways, vehicle-miles would seem to be the preferred weighting factor Use of other weighting factors could encourage improvements to low-volume roads simply to meet regulatory requirements, while higher volume roads go unattended. Deli neation of Travel Corridors Localities will require some guidance as they begin to delineate corridors or dist ricts within which levels of service are combined. This discussion will refer to such areas generically as transportation concurrency m anagement areas (TCMAs), a name coi ne d by Florida's state planning agency If TCMAs are too large, traffic problems will be glossed over and development decis i ons will be subject to challenge. We might expect property ow ne rs near the edges of large TCMAs, for example, to challenge p ro ject disapprovals prompt ed by traffic congestion at central locations or opposite edges If TCMAs are too small, flexibility to respond to systemwide needs will be sacrificed. In the extreme, TCMAs will cease to reflect motorists' experiences on typical trips or their 15


choice among alternative routes and simply become surrogates for individual facilities. For guidance in delineating TCMAs, the concepts of typical trips and alternate routes may be combined in the following general guideline: TCMAs should be drawn so as to encompass alternate routes available for common peak -hour trips. How the general guideline is operationalized is best left to local planners. Let it suffice to say that the guideline could be operationalized. For example, regional travel models could be used to generate tables of trip interchanges between traffic zones, and from these, common origin-destination pairs could be identified. Because level-of-service standards apply to peak hours, primary consideration might be given to work trip interchanges. Boundaries could be drawn so that traffic zones between which a majority of trip interchanges occur are part of the same TCMAs. IV. CHOICE OF ANALYSIS PERIOD Florida's administrative rules require that levels of service be analyzed for peak-hour conditions. Use of peak-hour volumes is consistent with standard engineering practice in facility design traffic operations, and traffic control. However, as Bill McShane and Roger Roess note in Traffic Engineering, .. .if peak hour volume is to be used as a common focus of design, operations, and control analyses, it is critical to understand which peak hour is being us ed "19 Among the multitude of choices are: 19 William R. McShane and Roger P Rocss, Traffic Engineering (1990), p. 63. 16


the single highest hour of the year, the 30th highest hour of the year, the lOOth highest hour of the year, the average peak hour of the peak season, or the annual average peak hour. Which peak hour is selected could have a dramatic effect on estimated levels of service. From Br evard County's Comprehensive Plan, traffic in the 30th highest hou r is 10% of annual average daily traffic = 0.10), traffic in the lOOth highest hour is 9% (K100 = 0.09), and traffic in the average peak hour is 8%. Thus, the traffic volume during the 30th highest hour is about 25% higher than the volume during the average peak hour; that extr a traffic could make as much as a four letter-grade difference in the estimated level of service. Interplay of Peak HouJ: and LOS Standards The choice of peak hour cannot be divorced from t he setting of l evel-of-servic e standards. The effect will be the same if a lower s tandard is applied to a higher-volume hour, or a higher standard is applied to a lower-volume hour. In its compreh ensive plan, Lee County adopted two standards -LOS D for the annual average peak hour and LOS E for the average peak hour of the peak season. The lo wer standard (LOS E) applied to t he peak season may be more restrictive tha n the hi g her standard (LOS D) applied to the entire year. Does this mean that there is no preferred peak hour for concurrency management 17


purposes? Not hardly, but it does mean that the choice of peak hour must be made on some basis other than the desire to promote or restrict development (which can be accomplished with any peak hour by simply lowering or raising level-of-service standards). Relevance of the Design Hour Customary practice in the United States is to base roadway design on the 30th highest hourly volume in the 20th year of service. This means that facilities are expected to operate at acceptable levels of service all but 29 hours of that year. Should the same peak hour be used for growth management and design purposes? The transportation planners interviewed for this study had mixed reactions. Some argued that their localities cannot afford to hold existing facilities to the same standard as new facilities. Others felt that the same standards must apply to new and existing facilities if growth management policies are to be internally consistent. The latter view appears more defensible than the former. Once a community has decided how much congestion it is willing to tolerate, and has set a level-of-service standard for a particular peak hour to implement its decision, that standard would logically apply to both new and existing roads. If it is considered acceptable for new roads to operate below the adopted standard for, say, 29 hours in the design year, it should be acceptable for existing roads to operate below the adopted standard for 29 hours in the current year, but no more than 29 hours. If jurisdictions cannot afford to maintain existing roads at the standard for new roads, 18


the solution is not to establish a lower standard of existing roads but rather to re-evaluate the standard for new roads. Standards for new roads are based on theoretical considerations and professional dictates, but funding availabiUty should be a factor, too. Indeed, it could be argued that the appropriate standard for new roads is that which is sustainable for existing roads, given funding availability. New and existing roads differ in one fundamental respect traffic volumes are known for existing roads and only estimated for new roads. This difference has been used to argue for higher standards in roadway design than growth manageme nt, where roadways are designed for a "margin of error" in traffic forecasts. If the result of underestimating future traffic volumes (congestion) is more serious than the result of overestimating them (wasted c apacity) it may be advantageous to make liberal assumptions in traffic forecasts. However, there is no reason for the choice of peak hour to reflect uncertainty in traffic forecasts, not when the forecasts themselves can be adjusted to reflect uncertainty. 30th Highest Hour Use of the 30th highest hourly volume in roadway design dates back to the 1950 Highway Capacity Manual. Traffic studies of that era had observed extreme variations in traffic flow on facilities from hour to hour, day to day, and season to season. When hourly traffic volumes for an entire year were graphed in order of descending magnitude, the resulting curves often dropped sharply at first and leveled off quickly. The 30th highest hourly volume was found to fall on the "knee" of the curves, where the slope changed 19


markedly.20 Based on this early work, it has become conventional wisdom that: The 30th highest hourly volume is the point of di. minishing returns in roadway design. It is uneconomical to design for volumes to the left of the 30th highest hour, since a great deal of capacity is required to meet demands that occur only a few times a year. It is shortsighted to design for volumes to the right of the 30th highest hour, since little additional capacity is required to accommodate demands that occur frequently. The "conventional wisdom" may be wrong in this case. Hourly traffic volumes in many localit ies do not follow the indicated pattern. Hourly volume curves tend to flatten out rather than remain static as areas become more developed; the knee of the c urve becomes a moving target, or disappears entirely. Even if hourly volume curves have predictable turning points, these points have no e conomic significance; the optimum design of a facility can only be determined by comparing the costs of alternative designs with the benefits to motorists.21 Plots of hourly traffic volumes for 20 represen tative FOOT permanent count stations in FY 1989 illustrate the a rbitrariness of the 30th highest hour (see examples in Figures 7). Several of the curves never level off and/or h ave no point at which the slope changes 20 U.S. Department of Commerce, Highway Caoacity Manual (1950), pp. 130-132. Martin Wohland Brian V Martin, Traffic System Analysis for Engineers and Planner< (1967), pp. 168. 20


dramatically. Even where there is a discernable "knee," it does not correspond to the 30th highest hour with any degree of consistency. The choice of design hour is ultimately a political rather than a technical matter. It involves balancing the public's desire to bold down road user taxes (which means more traffic congestion) against their desire to avoid traffic congestion (which means higher user taxes). lOOth Highest Hour? If statewide level-of-service standards are to have meaning, they must apply to the same peak hour throughout the state. FDOT has proposed a shift from the 30th highest to the tOOth highest hour as the basis for level-of-service determinations. Is this shift warranted and in the right direction? Use of the 30th highest hour ties level-of-service standards to the exceptional travel experience. It could be argued that standards should instead reflect the typical travel experience. The typical travel experience is represented by the annual average peak hour, o r more conservatively, by the average peak hou r of weekdays during the peak season. Table 5 presents average counts for these days at 20 FDOT permanent count stations. It also presents the 30th and lOOth highest hourly counts at these same stations For urban routes, the tOOth highest houdy counts are roughly equivalent to the average counts for peak hours of weekdays during the peak season.22 22 Recreational routes and some rural routes. are another matter. Due to weekend pcaki.og, tOOth highest hourly volumes often far exceed average peale hourly volumes for weekdays during the peale season. 21


The tOOth highest hourly volume would be easy to estimate, assuming this rough equivalence is borne out. I t would only be necessary to take one 24-hour count on a typical weekday during the peak season. The highest hourly count for that 24-hour period could be taken as an estimate of the tOOth highest hourly volume. This would improve on the practice in many localities of applying a generalized K-factor to a single, seasonally adjusted 24-hour traffic count Additionally, the tOOth highest hourly volume would be relatively easy to project Standard regional travel models forecast traffic volumes for the average weekday during the peak season. To obtain estimates of the lOOth highest hourly volume, it would only be necessary to apply a peak-to-daily ratio to model outputs. At present, modelers must first convert model outputs to annual average daily traffic volumes, and then a pply a generalized K-factor to the result. Peak Period Instead of Peak Hour? Daily peaks tend to spread out as urban areas grow and traffic congestion causes motorists to adjust their travel hours. Indeed, the largest cities do not have a "peak hour" per se but rather a two to three -hour period in th e morning and afternoon when commuting is heaviest. Roads become capacity-constrained and K-factors come to be determined by supply rather than demand. We can expect even more spreading of the peak as traffic congestion worsens and communities seek to better manage travel demand. With the state's approval, Dade County and the City of Miami based levels of service 22


There may be pages m1ss1ng from this document


Table 1 Roadway J..eVel0-SUvice Dererminations In LOcal Comprehensive Plans Pc:alr. Row-A.natyu:d. Brevatd County v/c R2tic (1965 HCM) Pan.llel Ro.adways 3CM Highest Hour Boca Raton FDOTTtbles None 301h Hi$hot Hour Oro'N2cd County FOOT Tables None 30th Highest tJour Charlou.e Col.lntv FOOI'Tt.blcs (1979 urPS ver.) Na<>e Avg, Pdit Hr. of Peak Coco< 'FOOT Table$ None 30th Highest Hou r CoUiet County FOOT T:ablcs None Avg. P<:,ak Ht. of Peak $uon Cond Springs Local (Based on LOS Pgm. None Not Specif.-ed DadeCounry FOOT Tables (1919 VTPS ver.) None Avg l H ighest Consec Wkdy Hrs Duval County FOOT Tables Nor.e 30th Highest HO\.'f F1asJcr County FDOTT11bl es and Roacf.Spcc:ifte None 30th Highest Hour Tables {Based on LOS Program) F1. 1.2vderdale FDOTTabtcs 2.0d Nooc Not Spectlicd An:aly.:;is wilh HCM Software Hillsborough Oy. FOOT Tables None 30th HJS}\C$1 Hour jcffe:son County FOOTTab!es None 30rridots CQmeeuo've Weekday Hour:s Orlando FOOT Arteri al Roads within TraiT:e lOOtli Hour Analysi.$ wi!hARTALL Pedomunee Oi$ttict.S Palm Ccy FOOT Tables; None 30th H ighest Hovr Pa.soo Coonry FOOT T2bles a.rtd Loca l None 30th .Ughest Hovr (Based on LOS Prog.n.m) Tampa Local Maxitnum Volume Tables Roads wilhin 30th Highdt Hour (B; d on 1965 HCM) 2nd Cirywide Vo lusia Counry FDO'T Tables No."'e Not Spec.if:cd Me:hocb c:iu::d are those used i.n.lhe adopted LCP; many i ocalities; Neve sir.ce to moresophi$ti_c;ated mMods


Table 2 Average Stopped Delay (in Seconds) Nonhbound AM Movements HCS ART-PlAN Kirkman Road Carrier to International 10.0 11.4 International to Major 7 3 7.3 Major to Vineland 9 8 Jl. l Vineland to Conroy 28. 8 Average 14.1 14.7 TurkeyUke Road Sand Lake to Wallace 3.4 3.4 Wallace to Panther 2.6 2 6 Panther to Paw 1.() 1.0 Paw to Hollywood 1.9 1.8 Hollywood to Production 1.2 1.2 Production to Vineland 0.9 0 8 Average 1. 8 1.8 S PEED STUDY 12 .15 6.0 ll.2 6.4 9.0 12.3 2.6 0 () 5.7 0 3.5


Table 3 Average Running Speed (in MPH) Northbound AM Movements HCS ART-PlAN Kirkman Road Carrier to lnternational 32.1 31.8 International to Major 368 46.7 Major to Vineland 33.2 33.1 Vineland to Conroy 35.2 35.3 Av-erage 35. 2 39.6 Turkey Lake Road Sand Lake to Wallace 38.9 38.9 Wallace to Panlher 38.2 38.2 Panther to Paw 32.4 32.4 Paw to Hollywood 29.8 29. 9 Hollywood to Production 30.5 30.6 Production tO Vineland 36 1 36.1 Average 36.6 36.7 SPEED SnJJ)Y 29.7 51.5 39.1 42.8 44.6 49. 6 46.2 48.4 30.8 29.0 38.9 44.7


Traffic Performance D istrict 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 5 T a b le 4 District Perform a n ce Criteria C i ty of Orl ando Percent of Lane Miles Operating at or Above Level of Service Standard 1995 2010 73% 75% 69% 73% 33% 5 2 % 32% 34% 79% 88% 81% 84% 88% 88% 80% 95% 60% 6 6 % 55% 50% 59% 62% 89% 97% 100% 100% 85% 91% 51% 58%


Table 5 Traffic Counts at Different Design Hours Station 30th lOOth Highest Peak Hour/Weekdays/ Hour Hour Peak Season 0117 1890 1736 1658 0087 2217 2069 2157 0166 1444 1345 1145 0013 2105 2049 2024 0161 4905 4815 4756 0096 2022 1958 2005 0145 594 548 554 0149 252 229 219 0038 2162 2048 2130 0118 1938 1748 1604 0047 879 811 644 0066 1913 1m 1594 0094 3494 3403 3419 0105 1424 1309 1326 0113 4571 4425 4481 0151 3494 3359 3421 0159 2222 2137 2231 0160 1076 989 948 0164 2167 2039 1697 0165 3350 3123 3320 Source: FOOT hourly traffic counts for F.Y. 1988-89.


PRE OS ION Figurel Alternative Methods of Arteri.:>J Analysis e Loc:al Muimum Volume T2bles e Generalized Maximum Volume T2bles Tr2vel Time 2nd Delay S

Figure 2 Average Travel Speeds and Levels of Service Kix'kman Road Average T r avel Speed (mph) Northbound AM Southbound Northbound Southbound PM HCS -ART-PLAN lwt>l Speed Study


Figure 3 Average Travel Speeds and Levels of Service l'ux"key Lake Road so.-2 Northbound Southbound Northbound Southbound AM PM HCS GilJSpeed Study


Figure 4 Estimated vs. Acrual Travel Speeds Broward Boulevard Average Travel Speed (mph) 30.------------------------------------------------, 2 5 -------..----.. ----------1 0 Eastbound Westbound Eastbound Westbound AM PM ART-P L AN -Speed Study


Figure 5 Regression Equation Relating Average Tra'Vel Speed to Peak Hour Traffic Volume and Signals Per Mile Seminole County Average Travel Speed 2 Lane Roads 44.7 R' 0.0087 :X (3.12) Standard Error Number of Observations Degrees of Freedom Peak Hour Traffte volume = 0.55 5.3 68 -65 t statistics shown in par entheses 7.74 (6.65) Signals Per Mile


----....... .. ......... j Figure6 T raffi c Volomes vs. Capacities (ny District) lee Cour)ty J.EGNO .A. HNU.l l. ft.AifiC VOLU'-111: OIIOWTH .ui...U.\. S.UI'neit YOWMC ... OWTM 'AOM llffVIliiJIU'oveut"Hra m o U INUAL CIOWTW JIII0\11 'OMMlTTI!O N,_flCVlW.IIIITI l*"'"'l"" f-----fWW5154t I*"'" =c--__ .. .I I I I I I I I I I I I I I I I I I I I I I I I I I I I ...... ""' ........ ..... _.,.., ,. ........... ........ ....... _...... ... --_...... .. -. ....... "'" ... --. ..... ........ ....... .-.--0 1S'I1UC'r I OIS11UC1' 2 PI OIS'rRIC'J' 5 DISTRICT 6 So urce: 1990 Amendm ent t o the P lan ( l .ec Co un ty Compre h ens i ve P l a n ) Volum e 1 o r 3 September 1990, 'l"ra rflc C i rculation l!xhlbil Vl-6, page VI-I 0. DISTRIC1"8


Figure 1 Urban Route Tallahassee US. 27 Station 151 12.0 -r-------------------------------------------------------------, f-. Q <(, 11.5 4. I J.l) ...... 0 .... s:: 11> () 10. 5 ... v P.. 11).0 -!).5 I IiI I I I I I'' j I' I I I I I I I' I' I I I I I I I I I I I I I j I I I'' I I I I j I' I'' I II I I I I' I I I I I I I I I I I I I I' I J I I I I I t I I I 0 20 10 GO GO 100 120 11 0 !GO IGO 200 Highest Hours of the Year


I 2. 0 I 1.5 -f-. o ll.o-1 -. I. -t ..... 0 ...., c

Figure 9 Recreational Route Palm Beach. SR. A-1-A Station 87 16 -.------------------------------------------------------------------, I : -r Cl 1 ..: 4. ..... 0 1:1 1 ...., .: Q) (.) .... (l)l2 o.. I I --10 () 20 40 60 00 100 120 140 160 100 200 Highest Hours of the Year