Transportation Research Record: Journal of the Transportation Research Board

The 2000 release of the Highway Capacity Manual (HCM) provides for the first time a corridor analysis method that guides users in the application of various chapters of the HCM to the analysis of automobiles and transit in a corridor. Together with the recent publication of the Transit Capacity and Quality of Service Manual (TCQSM), the HCM 2000 represents a significant advance in the direction of multimodal level-of-service (LOS) analysis. However, relatively little guidance is given in either the HCM or the TCQSM on the compilation of automobile and transit segment levels of service into a measure of corridor level of service. In addition, bicycles and pedestrians are ignored in the corridor methodology. A methodology was developed and tested in Florida for measuring and reporting the user-perceived quality of service for highway corridors from a multimodal perspective. Automobile and transit LOS analyses are based on the HCM 2000 and TCQSM, respectively. Bicycle and pedestrian levels of service are based on the bicycle and pedestrian LOS models, respectively. Four classes of corridors are recommended, and the methodology was tested on two classes of urban corridors, with and without a freeway. The methodology is applied in three steps: (a) corridor definition, (b) computation of modal level of service, and (c) reporting of results. The methodology was applied to six case studies throughout Florida at generalized and conceptual planning levels. Conclusions about the methodology were drawn from the case studies; the main conclusion is that the methodology provided a reliable overall indicator of corridor level of service by mode.

Interchanges have been the focus of several recent studies that attempted to identify the conditions in which one interchange type is more efficient than another. However, the guidance offered is of such a general nature that it is difficult to apply with confidence in a specific situation. The objective of this study was to describe an interchange evaluation technique that is of sufficient precision to facilitate interchange-type selection or planning-level evaluations of interchange performance. This technique was based on the use of characteristic curves that define the relationship between interchange delay and the sum-of-critical-flow-ratios. This sum represents a unique parameter that combines the effect of signal phase sequence, traffic volume, number of lanes, and saturation flow rate. A family of characteristic curves was developed and used to illustrate the interchange evaluation technique. The curves describe the relationship between delay and the sum-of-critical-flow-ratios for the single-point urban interchange (SPUI) and the tight urban diamond interchange (TUDI), both serving frontage-road systems. For the range of conditions considered, it was concluded that the TUDI will generally operate with less delay than the SPUI.

Development of a highway congestion index on the basis of expert and user opinions with fuzzy set theory was attempted. Forty 1-min videotape segments showing different highway traffic flow conditions were displayed to a group of highway experts and users. They were asked to complete the relevant questionnaires on the existence and acceptability of congestion for each video segment display. Five congestion-descriptive variables of travel speed, travel rate, delay rate, travel rate ratio, and delay ratio were determined for each video segment. Information about the values of congestion-descriptive variables for the displayed videos, questionnaire responses, and fuzzy set theory were used to develop congestion fuzzy set membership functions. The membership functions characterized the congestion intensity fuzzy values of “congested” and “unacceptably congested.” The membership functions were suggested as congestion indexes measured on a scale of 0 to 1. In the questionnaire, the surveyed group also provided opinions regarding the relative importance of the descriptive variables in explaining congestion. This information was used to determine average weighting values for the developed indexes. Using the average weighting values, overall indexes for “congested” and “unacceptably congested” were developed. The overall congestion indexes were also suggested for congestion intensity assessment and evaluation. The relationships between the Highway Capacity Manual levels of service and the developed congestion indexes were studied. The results were used to address some issues regarding normative index development. Although the study findings are based on a rather limited database and are location specific, the applied methodology can be used to capture congestion intensity perception in any future index development. The study index development was based on fuzzy set theory and is suggested as an alternative or complementary approach to past normative index development methods.

One of the issues involved in using microscopic simulation models is the variation in the simulation results. This study examined some of the more popular microscopic traffic simulation models, CORSIM, SimTraffic, and VISSIM, and investigated the variations in the performance measures generated by these models. The study focused on the capacity and delay estimates at a signalized intersection. The effects of link length, speed, and vehicle headway generation distribution were also investigated. With regard to variations in performance measures, the study found that CORSIM yields the lowest variations, whereas SimTraffic yields the highest. The highest variation in each simulation model normally occurs when the traffic demand approaches capacity. It was also found that delays are affected by the link length and speed in simulation models. Such an impact on delays is closely related to the range of speed variations. In general, shorter links and higher link speeds result in lower delays. There is no strong evidence that the headway distribution used to generate vehicles in the simulated network has any effect on capacity and delay estimates. Multiple simulation runs are necessary to achieve an accurate estimate on the true system performance measures. With a 10% error range in estimated delay, two to five runs may be enough for under-capacity conditions, but more than 40 multiple runs may be necessary to accurately estimate delay at, near, or over capacity.

Most existing analytical models determine signal settings at intersections by minimizing total delay. The main problem with these methods is that when green split is based on minimum total intersection delay, low-volume approaches are penalized. Although this may be appropriate, the extent of this penalty has never been theoretically resolved. A new concept and its methodology for determining traffic signal settings were proposed to provide a theoretical basis for this issue. This new concept, marginal delay, has its inception in marginal analysis and is defined as the increase in total delay resulting from a one-vehicle increase in the approach volume. Given the nature of the existing delay equations, marginal delay also represents the measure of the maximum individual vehicle delay for a given cycle. Three marginal delay equations were derived and analyzed over a range of 265 cases. Marginal delay analysis was applied to signal-setting calculations and compared with existing delay models. The work concluded that marginal delay is a significant variable in the calculation of signal settings, especially when large differences exist between the volumes of the different approaches. Furthermore, marginal delay used in conjunction with total delay can improve current signal-timing methods by quantifying the penalty applied to light-volume approaches and by providing a measure of variance between real-time flow and design flow.

The analyses conducted in this research were based on three methodologies for the field measurement of saturation headways. The first method (M1), the one on which most past studies were based, measured the characteristics of Vehicles 4 to 12 in a standing queue. M2, the method found in the Highway Capacity Manual (HCM), counted all vehicles in a standing queue, regardless of queue length. M3 included arrivals that joined the standing queue as long as vehicles were up to 140 ft from the stop line. This study focused on one approach of a high-design intersection with heavy, random arrivals. The large number of observations and the practically ideal traffic conditions enabled the acquisition of several statistically significant results on saturation flow (s), start-up lost time (SULT), and start-up response time (SRT): (a) when long queues are present, the typical field measurement of s based on the first 12 vehicles is an overestimate of s for through vehicles and an underestimate of s for protected left-turning vehicles; (b) the type of movement had a more dominant role in determining s than the level of saturation (or queue length); (c) SRT displayed a bigger variation than headways— the left-turning movement had a significantly shorter SRT than the through movement did; and (d) much higher SULTs were estimated in this study compared with those in the HCM.

An iterative model for computing capacities at all-way stop-controlled (AWSC) intersections has been included in the new Highway Capacity Manual (HCM) 2000. The model is based on five saturation headway values, each reflecting a different level of conflict faced by the subject-approach driver. From this model the capacity and service time at any approach can be computed using iterative calculations. The model in the HCM is a so-called approach-based model, which only takes into account the conflicting cases among the approaches. The effect of turning streams or movements is not modeled in sufficient detail. In contrast, a theoretical, stream-based model for determination of capacities at AWSC intersections has been developed. This model is based on the addition-conflict-flow (ACF) method developed from graph theory and takes into account all the traffic streams at the intersection. This allows a systematic and realistic analysis of the traffic process at AWSC intersections. The computational procedure included in the model can be conducted without iterative calculation steps. The ACF and the HCM models for intersections with single-lane approaches were comprehensively validated, as well as a modified version of the HCM model that significantly enhances its features. The results of the validation indicate that the total capacity of an AWSC intersection with single-lane approaches based on the HCM model ranges between 1,450 and 1,550 passenger cars per hour (pc/h), whereas the total capacity based on the ACF model ranges between 1,600 and 2,000 pc/h. The modified HCM model yields total capacities ranging between 1,700 and 2,000 pc/h. The ACF model and the modified HCM model yield similar capacity results under normal traffic flow conditions.

A planning-level adaptation was developed of the Highway Capacity Manual (HCM) 2000 procedure for estimating the level of service (LOS) on two-lane and multilane highways in Florida. The problems associated with planning-level adaptations in general and with uninterruptedflow highways in particular were identified. Although much of the adaptation was achieved though the use of default values for data items, some departures from the HCM procedures were required. The most significant deviation was the creation of a third class of two-lane highway to supplement the two classes currently defined in the HCM. A case was made for the existence of this class and its inclusion in a future edition of the HCM. The Florida Department of Transportation’s planning-level methodology, termed HIGHPLAN, is well suited to its intended application, which is planning-level analysis of two-lane and multilane highways in Florida. It maintains fidelity to the HCM procedures to the extent that Florida conditions will allow and Florida users will accept. As long as they are understood, the departures from the HCM should not pose significant problems for users outside of Florida. The planning-level methodology has also been implemented in a software program that produces LOS estimates and service volume tables covering sitespecific conditions.

A comparison of delay on arterials was made using simulation and Highway Capacity Manual (HCM) computations. An extensive set of cases was defined over a range of flow rates, free-flow speeds, and signals per mile. Delay was calculated on the basis of the HCM computation for the specified signal progressions. The same cases were run using a specially revised version of the NETSIM microscopic simulation model. Detailed analysis showed that the discrepancies were traceable to the progression factor (PF). Corrections can be made (and are outlined) but must be done with care. There are a number of complex mechanisms at work. The primary result of this study may be clarifying the insufficiency in the current PF algorithm. The expectation that a single multiplicative factor could provide the needed correction to a long-link random-arrival steady-state delay model (that is, the existing HCM delay model), so that it could be a network-based model for delay estimation, appears to be simply unrealistic. For these and other reasons, research is focusing on the creation of a time-varying intersection delay model, building from basic principles.

Many measures have been proposed to represent the status of traffic conditions on arterial roadways in urban areas. The debate about what is the most appropriate measure continues. In a contribution to the debate, another approach was offered. Traditionally, two general approaches exist. One is based on the relationship between supply and demand. The other is a measure relative to the most acceptable status of service quality. The latter measure allows the public to relate to their travel experience. In either case, however, derivation of measures of congestion involves uncertainty because of imprecision of the measurement, the traveler’s perception of acceptability, variation in sample data, and the analyst’s uncertainty about causal relations. A measure is proposed that is a composite of two traditional measures, travel speed and delay. In recognition of the uncertainty, a fuzzy inference process was proposed. The inputs are travel speed, free-flow speed, and the proportion of very low speed in the total travel time. These values were processed through fuzzyrule-based inference. The outcome was a single congestion index value between 0 and 1, where 0 is the best condition and 1 is the worst condition. The process was demonstrated using real-world data. The results were compared with those of the Highway Capacity Manual. Although no conclusion can be drawn about the best measure of congestion, the proposed inference process allows the mechanism to combine different measures and also to incorporate the uncertainty in the individual measures so that the composite picture of congestion can be reproduced.

A proposed macroscopic methodology for measuring the level of service (LOS) of toll plazas has been developed using delay as the measure of effectiveness (MOE). On the basis of field research and data analyses, the 85th percentile of the cumulative individual vehicular delay was found to be the most comprehensive measure for evaluating the LOS at a toll plaza. Other MOEs were examined but found to be less flexible with different plaza configurations and lane payment types. More than 55,000 individual vehicular records from three mainline toll plazas in Orlando, Florida, representing eight different plaza configurations with varied percentages of electronic toll collection (ETC) usage were used to validate the methodology. TPSIM, a toll plaza simulation model, was used to produce an additional 49 scenarios representing the three plazas with varied percentages of ETC usage and 21 additional plaza configurations. Service time was examined to determine the level at which a driver begins to feel discomfort and inconvenience at a toll plaza. An LOS hierarchy was established based on the conclusions of this analysis, feedback from professionals, and reference to the Highway Capacity Manual (HCM) 2000. The 85th delay percentile graphs from each of the plaza analysis results for LOS values were also observed to be similar.

The Highway Capacity Manual (HCM) presents a procedure for estimating signalized intersection control delay, which is used to determine level of service (LOS) and to evaluate intersection performance. The HCM is used extensively by traffic engineers. However, it is intended as an offline decision support tool for planning and design. To meet user requirements of advanced traffic management and information systems, new LOS criteria are required for real-time intersection analysis. The objective of this research was to demonstrate a technique for development of such LOS criteria. The study used a new measure of effectiveness, called reidentification delay (RD), derived from analysis of vehicle inductive signatures and reidentification of vehicles traveling through a major signalized intersection in the city of Irvine, California. Two main issues regarding real-time LOS criteria were tackled. The first was how to determine the threshold values partitioning the LOS categories. To provide reliable real-time traffic information, the threshold values should be decided on so that RDs within the same LOS category would represent similar traffic conditions as much as possible. RDs in different LOS categories should also represent dissimilar traffic conditions. The second issue concerned the aggregation interval to use for RD in deriving LOS categories. An investigation of both fixed and cycle-based aggregation intervals was conducted. Several clustering techniques were then employed to derive LOS categories, including k-means, fuzzy, and self-organizing map approaches. The resulting real-time LOS criteria were then determined. The procedures used in this study are readily transferable to other signalized intersections for the derivation of real-time LOS.

The Highway Capacity Manual (HCM) 2000 provides methods to estimate performance measures and the level of service for different types of traffic facilities. Because neither the input data nor the model parameters are totally accurate, there is an element of uncertainty in the results. An analytical method was used to estimate the uncertainty in the service measures of two-lane highways. The input data and the model parameters were considered as random variables. The propagation of error through the arithmetic operations in the HCM 2000 methodology was estimated. Finally, the uncertainty in the average travel speed and percent time spent following was analyzed, and four approaches were considered to deal with uncertainty in the level of service.

The operation of freeway weaving sections is characterized by intense lane-changing maneuvers and complex vehicle interactions that often create bottlenecks along freeway facilities. The CORSIM microscopic simulation model was applied to simulate the operation of eight realworld weaving sites in California under a wide range of operating conditions. The results indicate that CORSIM with default parameter values underpredicts the speeds in the weaving section by about 19% on average. Numerous simulation runs were made with different values of the model parameters. The following parameters were found to significantly affect the CORSIM results: (a) car-following sensitivity factor, (b) lane-changing aggressiveness factor, and (c) percentage of freeway through vehicles that yield to merging traffic. The calibrated CORSIM model reasonably replicated observed traffic operations at all test sites. The predicted average speeds were within ±5 mph for most test sites. Good agreement between measured and predicted values was obtained for all the combinations of design characteristics and demand patterns.

Highway Capacity Manual (HCM) 2000 two-lane rural highway directional, two-way, and passing-lane analysis procedures based on field data and simulation were evaluated. Detailed field data were collected from two highway sections in northern Idaho, one with and one without passing lanes. The TWOPAS simulation model was used to provide additional insights. Particular attention was given to the differences in estimates for percent time spent following (PTSF) produced by the twoway and directional analysis procedures. It was found that the two-way analysis procedure was more accurate, although both procedures produced estimates that were too high. The passing-lane analysis procedure was also evaluated, and the HCM 2000 procedure was found to be conservative in its estimates of PTSF reductions due to a passing lane.

The delays for signalized intersections from the Highway Capacity Manual [using the Highway Capacity Software (HCS)], PASSER II-90, PASSER IV-96, Synchro, and CORSIM software were compared. An examination was made of how much delay reduction materialized when optimized signal settings were implemented. Synchro should not be compared with HCS unless certain precautions are taken to make the conditions analyzed in Synchro comparable with those analyzed in HCS. For both pretimed and actuated signals, if simultaneous phases exist in which one of them terminates ahead of the others, the results of HCS and Synchro are not comparable. For both pretimed and actuated signals, when the phases terminate simultaneously, whether the data are imported from Synchro or manually entered in HCS, the condition analyzed will not be comparable unless precautions are taken. Once these precautions were taken, control delays from Synchro and HCS were not significantly different for pretimed, uncoordinated signals. For actuated—uncoordinated and for actuated—coordinated controllers, the comparisons are not meaningful. Delays for optimized conditions for PASSER II and PASSER IV were not significantly different from the delays before optimization. However, for Synchro, the delays for optimized conditions were significantly lower than the delays before optimization. CORSIM was used to assess the impacts of the optimization. If optimized signal settings suggested by PASSER II, PASSER IV, or Synchro are implemented, the delays for the optimized condition would not be significantly lower than the delays before optimization.

An empirical study was undertaken of congested patterns at highway bottlenecks. On the basis of statistical data it was found that the spatialtemporal structure of congested patterns possesses some predictable features. From these features a classification of congested patterns was made. It was found that the most frequently observed congested pattern is the general pattern (GP). In GP synchronized flow occurs upstream of a bottleneck and wide moving jams spontaneously emerge in that synchronized flow. Capacity in free flow can be about twice as high as capacity in congested traffic upstream of the on-ramp if the GP has formed.

In previous studies, two traffic data-cleaning algorithms were developed at the Institut National de Recherche sur les Transports on the basis of filtering techniques and statistical approaches. Because of their mathematical structure (linearity of the process), both algorithms present a high level of inaccuracy in the case of nonhomogeneous traffic conditions at the location of the measurement stations (for example, free flow upstream and congestion downstream, or vice versa). A new algorithm for solving the traffic data-cleaning problem on the basis of real-time application of a dynamic first-order modeling approach was devised to take into account the nonlinearity of the traffic phenomenon. The developed algorithm, named PROPAGE, was tested using real data measurements, including a wide spectrum of traffic conditions. Compared with results from previous algorithms, the results obtained were more accurate.

The real-time kinematic differential Global Positioning System (GPS) has facilitated a new horizon in traffic engineering. Multiple car-following experiments conducted with a real-time kinematic GPS with 10 vehicles participating in a probing field gave high-quality results in headway, speed, relative speed, and acceleration. The expected accuracies for measuring position and speed were 10 mm and 0.16 km/h, respectively. The vehicles were driven in a loop consisting of two parallel straight sections connected by two semicircular curves. Different driving conditions were induced in the platoon by instructing the leading driver to follow predetermined speed variations. The experiments yielded sets of continuous observations. Headway, speed, and acceleration were measured using conventional equipment for the purpose of comparing accuracy. The accuracy of the data obtained using the GPS was superior to that of the same data obtained using conventional measurements. The variation in driving characteristics down the stream of vehicles was studied using the experimental data. The results showed that the reaction time between a change in relative speed and the corresponding change in acceleration varies during the driving process. The reaction time of individual drivers also changes along the platoon. The good-quality data were able to give high-resolution plots of acceleration and relative speed illustrating that both the reaction time and the functional relationship between acceleration and relative speed do not remain constant.

A methodology was developed to find appropriate travel times for highway links using data from point detectors that could be at various points within the link or could even be outside the link. The travel times were found using a definition that the appropriate value is the one experienced by a virtual vehicle reaching the midpoint of the link at the midpoint of the time step. A simple iterative scheme was proposed to find the travel time profiles. The accuracy of the scheme depends on whether aggregated detector data or individual vehicle spot speeds are used. Comparison of estimated travel times with actual experienced travel times of all vehicles in a microscopic simulation showed the technique to give very good results, comparable with having a high number of probe vehicles reporting travel times.

A hybrid neuro-fuzzy application for short-term freeway traffic volume forecasting was developed. The hybrid model consists of two components: a fuzzy C-means (FCM) method, which classifies traffic flow patterns into a couple of clusters, and a radial-basis-function (RBF) neural network, which develops forecasting models associated with each cluster. The new hybrid model was compared with previously developed clustering-based RBF models. In addition, the dynamic linear model was studied for comparison. The study results showed that the clustering-based hybrid method did not produce time-lag phenomena, whereas the dynamic linear model and the RBF model without clustering revealed apparent time-lag phenomena. The forecasting performance for freeway traffic volumes from the San Antonio, Texas, TransGuide system shows that even though the hybrid of the FCM and RBF models appears to be promising, additional research efforts should be devoted to achieving more reliable traffic forecasting.

The Lighthill-Whitham-Richards kinematic wave traffic flow model was extended to describe traffic with different types of vehicles, in which all types of vehicles are completely mixed and travel at the same group velocity. A study of such a model with two vehicle classes (e.g., passenger cars and trucks) showed that when both classes of traffic have identical freeflow speeds, the model (a) satisfies the first-in-first-out rule, (b) is anisotropic, and (c) has the usual shock and expansion waves and a family of contact waves. Different compositions of vehicle classes in this model propagate along contact waves. Such models can be used to study traffic evolution on long crowded highways where low-performance vehicles entrap high-performance ones.

The queue discharge problem at a signalized intersection was analyzed with application of the modified Pitt car-following system. The Pitt car-following system was implemented on an Excel spreadsheet in the form of a prototype simulation model. The situation consisted of a single intersection in which discharging vehicles were unconstrained by downstream conditions. It was asserted that the car-following parameters used in the queue discharge problem are significantly different from those used in the uninterrupted-flow problem and that this is consistent with the two-fluid model. A number of scenarios were executed for each situation, using the spreadsheet implementation to examine a number of issues, including the effect of the following: free-flow speed, car-following parameter variations, vehicle length, heterogeneous traffic streams, and lane changing. It was shown that all of these issues have significant effects on discharge headways and that the Pitt car-following system has a far wider range of applicability than the simple equal headway model. The intuitiveness of the findings and calibration and validation issues were also addressed.

A microscopic traffic flow model based on the constant-time-headway policy and McRuer’s man-machine crossover model was designed. Automatic control theory concepts were employed in the model formulation. The constant-time-headway policy was used to generate the command model of a human driver’s decision for vehicle acceleration or deceleration. This command is the input signal fed into the driver-vehicle dynamics suggested by the crossover model. The proposed model was mathematically formulated, designed, implemented, and numerically simulated. The stability properties and validity of the proposed model were analyzed on the basis of the simulation results. It was demonstrated that the proposed model can reproduce well-known traffic phenomena such as shock waves, intersection starting and stopping waves, and loop structures of flow-density and speed-density plots.

In a review of the empirical literature on congested freeway flow, with special emphasis on topics of current interest, including congested flow phases, congested-regime flow-concentration relationships, wave speeds, average time-gap behavior, and transitions between uncongested and congested flow, the most striking feature of the results is their diversity. Shapes of flow-concentration relationships vary among studies and sites, wave speeds vary, patterns of repetition in congested flow appear to vary widely, and flow breakdown occurs at a variety of locations and for a variety of reasons. The diversity of results seems to indicate a similar diversity of phenomena rather than the effect of methodological differences. For the most part, methodologies are simple and intuitive, and because of their simplicity, they are unlikely to result in variability of the magnitude found in the results. Several methodological features that could possibly distort results are noted but do not appear to be responsible for the diversity of results. A major limitation of research to date is a lack of systematic understanding of the relative frequency of occurrence of different congested-flow phenomena. Because abundant loop-detector data are now available, research into the frequencies of congested-flow phenomena is feasible; however, analysis of very large quantities of data will be required.

A model of disaggregate interactions between individual vehicles was developed that makes explicit the vehicle classes, the class trip rates, and their kinematic parameters (free speed, acceleration rate, length, safety margin). Assuming exponential gaps between vehicles, analytic formulas were derived for the mean value and the variance of the path travel times within each class. The model was successively applied to one-lane roads; two-way, two-lane roads; one-way, two-lane roads; and two-way, three-lane roads.

A concept of a car-following system that consists of a car-following equation together with a set of constraints is appropriate for micro-simulation models. Recognition was given to the asynchronous condition in the leader-follower relationship as it relates to the internal dynamics in simulation models. The Pitt car-following equation and criticisms of it were evaluated. Two modifications to the Pitt car-following equation were introduced that address the criticisms. The first modification consists of a change in the method by which the perception-reaction time is handled, eliminating the requirement that it must be much less than the simulation time step and removing a discrepancy between the application of car-following and emergency deceleration. The second modification consists of a multiplicative factor, K, that addresses issues related to overdamping and the dead band, which have been raised concerning the original formulation of the Pitt model. An Excel spreadsheet was used to implement the car-following system, and a series of scenarios were simulated. The results of the scenarios indicated that the addition of the factor K has a substantial effect on the behavior of vehicles in a platoon when the leader of the platoon is subjected to a deceleration. Issues of calibration and validation were also addressed.

Three car-following models were compared: the Greenshields single-regime model, the Pipes two-regime model, and a four-parameter single-regime model that amalgamates both the Greenshields and Pipes models. The four-parameter model proposed by Van Aerde and Rakha is less known but is currently implemented in the INTEGRATION 2.30 software. The Greenshields and Pipes models were considered because they represent state-of-the-practice models for several types of microscopic and macroscopic software. The Greenshields model is widely used in macroscopic transportation planning models. In addition, the Pipes model is implemented in a number of microscopic traffic simulation models including CORSIM and VISSIM. Steady-state car-following behavior is also related to macroscopic traffic stream models to develop calibration procedures that can be achieved using macroscopic loop detector data. The study concluded that the additional degree of freedom that results from including a fourth parameter (Van Aerde model) overcomes the shortcomings of the current state-of-the-practice traffic stream models by capturing both macroscopic and microscopic steady-state traffic behavior for a wide range of roadway facilities and traffic conditions. Also developed was a procedure for calibrating the Pipes car-following model using macroscopic field measurements that can be obtained from loop detectors. Although this calibration procedure does not overcome the inherent shortcomings of the Pipes model, it does provide an opportunity to calibrate the CORSIM and VISSIM car-following behavior to existing roadway conditions more efficiently and without the need to collect microscopic traffic data.

An optimization framework for online flow propagation adjustment in a freeway context was proposed. Instead of performing local adjustment for individual links separately, the proposed framework considers the interconnectivity of links in a traffic network. In particular, dynamic behavior in the mesoscopic simulation is approximated by the finite-difference method at a macroscopic level. The proposed model seeks to minimize the deviation between simulated density and anticipated density. By taking advantage of the serial structure of a freeway, an efficient dynamic programming algorithm has been developed and tested. The experiment results compared with analytic results as the base case showed the superior performance of dynamic programming methods over the classical proportion control method. The effect of varying update intervals was also examined. The simulation results suggest that a greedy method considering the impact of inconsistency propagation achieves the best trade-off in terms of computation effort and solution quality.