6.0 Comparison of Highway Capacity Manual (HCM) and Simulation

The intent of this section is to compare the capabilities of HCM and traffic simulation tools and to provide additional guidance on assessing when traffic simulation may be more appropriate than HCM-based methods or tools.

6.1 HCM Strengths and Limitations

For many applications, HCM is the most widely used and accepted traffic analysis technique in the United States. The HCM procedures are good for analyzing the performance of isolated facilities with relatively moderate congestion problems. These procedures are quick and reliable for predicting whether a facility will be operating above or below capacity, and they have been well tested through significant field-validation efforts. However, the HCM procedures are generally limited in their ability to evaluate system effects.

Most of the HCM methods and models assume that the operation of one intersection or road segment is not adversely affected by conditions on the adjacent roadway. Long queues at one location that interfere with another location would violate this assumption. The HCM procedures are of limited value in analyzing queues and the effects of the queues.

There are also several gaps in the HCM procedures. HCM is a constantly evolving and expanding set of analytical tools and, consequently, there are still many real-world situations for which HCM does not yet have a recommended analytical procedure. The following list identifies some of these gaps:

• Multilane or two-lane rural roads where traffic signals or stop signs significantly impact capacity and/or operations.
• Climbing lanes for trucks.
• Short through-lane is added or dropped at a signal.
• Two-way left-turn lanes.
• Roundabouts of more than a single lane.
• Tight diamond interchanges.

6.2 Simulation Strengths and Limitations

Simulation tools are effective in evaluating the dynamic evolution of traffic congestion problems on transportation systems. By dividing the analysis period into time slices, a simulation model can evaluate the buildup, dissipation, and duration of traffic congestion. By evaluating systems of facilities, simulation models can evaluate the interference that occurs when congestion builds up at one location and impacts the capacity of another location. Also, traffic simulators can model the variability in driver/vehicle characteristics.

Simulation tools, however, require a plethora of input data, considerable error checking of the data, and manipulation of a large amount of potential calibration parameters. Simulation models cannot be applied to a specific facility without the calibration of those parameters to the actual conditions in the field. Calibration can be a complex and time-consuming process.

The algorithms of simulation models are mostly developed independently and are not subject to peer review and acceptance in the professional community. There is no national consensus on the appropriateness of a simulation approach.

Simulation models, for all their complexity, also have limitations. Commercially available simulation models are not designed to model the following:

• Two-way left-turn lanes.
• Impact of driveway access: Major driveways can be modeled as unsignalized T-intersections. However, models cannot address the impact of numerous minor driveways along a street segment (link). They can only be approximately modeled as a midblock sink or node.
• Impact of onstreet parking, commercial vehicle loading, and double parking (although such conditions may be approximately modeled as short-term incidents).
• Interferences that can occur between bicycles, pedestrians, and vehicles sharing the same roadway.

Simulation models also assume “100-percent safe driving,” so they will not be effective in predicting how changes in design might influence the probability of collisions. In addition, simulation models do not take into consideration how changes in the roadside environment (outside of the traveled way) affect driver behavior within the traveled way (e.g., obstruction of visibility, roadside distractions such as a stalled vehicle, etc.).

6.3 Traffic Performance Measures: Differences Between HCM and Simulation

The HCM methodologies and tool procedures take a static approach to predicting traffic performance; simulation models take a dynamic approach. HCM estimates the average density, speed, or delay over the peak 15 minutes of an hour, while simulation models predict density, speed, and delay for each time slice within the analysis period (which can be longer than an hour).

Not only are there differences in approach, there are differences in the definitions of the performance measures produced by the simulation models and the HCM tools. Some of the most notable differences include:

• Simulation models report density for actual vehicles, while HCM reports density in terms of equivalent passenger cars (trucks and other heavy vehicles are counted more than once in the computation of density).
• Simulation models report vehicle flows in terms of actual vehicles, while HCM reports capacity for freeways and highways in terms of passenger-car equivalents.
• Simulation models report delay only on the street segment where vehicles are slowed down, while HCM reports all delay caused by a given bottleneck (regardless of the actual physical location of the vehicles).
• Simulation models report queues only on the street segment where vehicles are actually queued, while HCM reports all queued vehicles caused by a given bottleneck (regardless of the actual physical location of the vehicles).
• Simulation models do not necessarily report control delay at signalized intersections. The reported values include midblock delays for vehicles traveling along the link, or only stopped delay at the traffic signal.

6.4 Strategy for Overcoming the Limitations of HCM

After a transportation professional has decided that the HCM procedures do not meet the needs of the analysis, the next step is to determine whether microscopic, mesoscopic, or macroscopic simulation is required. There are several simulation programs available for evaluating a variety of transportation improvements, facilities, modes, traveler responses, and performance measures. These analytical tools vary in their data requirements, capabilities, methodology, and output. In addition, the performance measures for the simulation models and the HCM procedures may differ in definition and/or methodology (e.g., the number of stops may be estimated at speeds of less than 8 kilometers per hour (km/h) (5 miles per hour (mi/h)) for one tool, but 0 km/h for another).

If it is not necessary to microscopically trace individual vehicle movements, then the analyst can take advantage of the simpler data entry and control optimization features available in many macroscopic simulation models. However, macroscopic models often have to make certain assumptions regarding regularity in order to be able to apply macroscopic vehicle behavior relationships. If these assumptions are not valid for the situation being studied, then the analyst must resort to mesoscopic or microscopic simulation.

Simulation models require a considerable amount of detailed data for input, calibration, and validation. In general, microscopic simulation models have more demanding data requirements than mesoscopic and macroscopic models. Simulation models are also more complicated and require a considerable amount of effort to gain an understanding of the assumptions, parameters, and methodologies involved in the analysis. The lack of understanding of these tools often makes credibility and past performance (use/popularity) a major factor in the selection of a particular simulation tool. More information on this issue may be found in Guidelines for Applying Traffic Microsimulation Modeling Software (Volume III).

Table of Contents | List of Tables | List of Figures | Top of Section | Previous Section | Next Section | HOME