Appendix B: CORSIM Capabilities and Limitations
This appendix identifies some of the capabilities and limitations of the CORSIM model.
CORSIM Capabilities
CORSIM can model many elements of modern traffic systems explicitly. Many of the other features can be approximated using the basic elements of CORSIM, along with engineering judgment. There are other tools that may be better suited for a given situation. This section is to be used to decide if CORSIM is the appropriate tool. The authors do not want to promote the use of CORSIM for situations in which we know there are tools that may be better suited for the situation.
There may be modeling situations where CORSIM models some portion of the network better than other tools but the other tools may model different portions of the network better than CORSIM. Only the analyst’s engineering judgment can be used to decide which tool to use. For example, if the sole purpose of the analysis is to model roundabouts, CORSIM is not the tool to use. If the main purpose of the analysis is to model a freeway segment that has a roundabout at the end of one of the off-ramps, CORSIM may be a viable option.
Table 21 and Table 22 are arranged into freeway and surface street systems and show CORSIM built-in capabilities.
Table 21. CORSIM freeway modeling capabilities.
Freeway Modeling Features |
CORSIM Capability? |
---|---|
Bottlenecks |
Yes |
HOV lanes, including ramp meter bypass |
Yes |
Barriers |
Yes |
Lane closures |
Yes |
Interchanges |
Yes |
Weaving sections |
Yes |
Toll booths / weigh stations / draw bridges |
No* |
Oversaturation (congestion) |
Yes |
Time-varying demand |
Yes |
Trucks biased or restricted to specific lanes |
Yes |
Incidents |
Yes |
Workzones |
Yes |
Surveillance |
Yes |
Fixed-time ramp metering |
Yes |
Time varying fixed-time ramp metering |
Yes |
Traffic-responsive ramp metering |
Yes |
Bus operations |
Yes |
*These features are not modeled explicitly by CORSIM, but can be modeled using basic CORSIM elements. Some of them require using both surface street elements and freeway elements.
Table 22. CORSIM surface street modeling capabilities.
Surface Street Modeling Features |
CORSIM Capability? |
---|---|
Bottlenecks |
Yes |
Complex intersections |
Yes |
HOV lanes |
Yes |
On-street parking |
Yes |
Two-way left turn lanes |
No* |
Oversaturation (congestion) |
Yes |
Time-varying demand |
Yes |
Incidents |
Yes |
Delay due to Pedestrians |
Yes |
Workzones |
Yes |
Unsignalized intersections |
Yes |
All-way stops |
Yes |
Roundabouts |
No* |
Fixed-time signals |
Yes |
Actuated signals |
Yes |
Signal coordination |
Yes |
Surveillance |
Yes |
U-turns |
No* |
Transit signal priority |
No |
Bus operations |
Yes |
Light rail |
No* |
*These features are not modeled explicitly by CORSIM, but can be modeled using basic CORSIM elements. Some of them require using both surface street elements and freeway elements. Appendix K contains some examples of ways to code situations that are not directly modeled by CORSIM.
CORSIM Limitations
Like all models, CORSIM is an approximation of the real world. CORSIM has limitations that must be understood in order to use the model correctly. Understanding the limitations will also prevent the analyst, reviewer, decision maker, and reviewing public from expecting too much from the model. Some of these limitations may warrant using a different simulation tool.
The components of CORSIM were designed as flow models to generate the correct overall flow on the links. New users of CORSIM often expect it to model the smallest details of driver behavior and vehicle interaction perfectly. During CORSIM’s development, individual vehicle interaction details were not as critical as the overall traffic flow. CORSIM was designed to operate very fast and efficiently. When the vehicle animation capabilities were added, many vehicle interaction problems were discovered. Many problems have been fixed; however, some low-level vehicle interaction problems still exist. Scrutinizing vehicle animation will usually yield a situation that is not exactly life-like. The analyst and reviewers must emphasize that the calibrated model will match the real world traffic flows, which is what CORSIM was designed to do.
One reason some driver behaviors and vehicle interactions do not match the real world is, by default, CORSIM uses a one-second time step. Vehicle positions and interactions are only updated every whole second. In the real world, drivers begin to react and apply accelerations and decelerations in smaller time frames than one second. Beginning with TSIS/CORSIM version 6.0 the freeways can be updated at less than one second intervals, although the sub-second movements cannot be animated in TRAFVU. In many cases, this improves the vehicle interaction but it slows the execution time of CORSIM.
CORSIM is a one-dimensional model augmented at lane change. Each vehicle’s actions are based on the current link and current lane only. For safety reasons, real-world drivers often adjust their speed to reduce the speed differential between lanes. In CORSIM, vehicles do not adjust their speed based on vehicles in adjacent lanes unless a lane change is required.
Each vehicle follows the vehicle directly in front of it and in most cases is only aware of that vehicle. In the real world, drivers often react to the actions of the vehicles downstream of the leader. CORSIM does not have this capability. A CORSIM vehicle does not have the capability of making lane choice based on the type of vehicles downstream. A real-world driver may change lanes to avoid being behind a large slow moving vehicle that is currently stopped in queue at a traffic signal. CORSIM does not have this look-ahead decision making process.
No major adjustments are made by CORSIM vehicles in order to find a better gap. Real-world drivers will speed up or slow down in response to larger gaps ahead of them or behind them. CORSIM vehicles will make adjustments to find a better gap based only on the vehicles directly adjacent to them. Real-world drivers on an on-ramp may adjust their speeds upstream of the gore point to fit into a gap in the mainline traffic. CORSIM vehicles do not have this capability.
CORSIM vehicle positions are based on the link they are on, the lane they are in, and the distance from the upstream end of the link. All vehicles are in a lane at all times. There is no capability to move a vehicle over by half of a lane. At the point where lanes are added or dropped there is no transition region. The lane is either there or it is not there. TRAFVU depicts a transition region (lane taper) for display purposes only.
CORSIM surface streets can only have seven lanes of traffic total. There can be only two turn bays on each side of the link. There are no acceleration lanes on surface streets. CORSIM freeway links can have five mainline lanes plus three auxiliary lanes on the left side of the link and three auxiliary lanes on the right side of the link. However, full auxiliary lanes may be used as through lanes in most situations. This allows a total of 11 through lanes.
Internal nodes can be labeled from 1 through 6999, limiting the number of internal node to 6,999 nodes. Interface nodes can be labeled from 7000 through 7999, limiting the number of interface nodes to 1,000 nodes. Entry and exit nodes can be labeled from 8000 through 8999, limiting the number of entry/exit nodes to 1,000.
The number of links is only limited by the number of nodes and the number of connections allowed at a node. The maximum number of approaches to a surface street node is five. The maximum number of approaches or departures from a freeway node is two.
The maximum length of a surface street is 3,047.7 m (9,999 ft). The minimum length of a surface street is 15.2 m (50 ft). Using all possible nodes and two direction surface links and the maximum link length, it would be theoretically possible to model over 41,842.9 km (26,000 mi) of surface streets in one network.
The maximum length of a freeway link is 30,479.7 m (99,999 ft). The minimum length of a freeway link is dependent on the maximum distance the fastest driver type can travel in one second based on the free-flow speed. Normally the minimum length is around 33.5 m (110 ft). Ramp links or freeway links with low free-flow speeds may have much shorter lengths. Using all possible nodes and the maximum link length, it would be theoretically possible to model over 212,433.4 km (132,000 mi) of freeway in one network.
The maximum length of a freeway link is 30,479.7 m (99,999 ft). The minimum length of a freeway link is dependent on the maximum distance the fastest driver type can travel in one second based on the free-flow speed. Normally the minimum length is around 33.5 m (110 ft). Ramp links or freeway links with low free-flow speeds may have much shorter lengths. Using all possible nodes and the maximum link length, it would be theoretically possible to model over 212,433.4 km (132,000 mi) of freeway in one network.
Freeway links have a maximum free-flow speed of 112.6 kilometers per mile (km/h) (70 miles per hour (mi/h)). This is the maximum mean speed that can be input to the model. Some vehicles with more aggressive driver types will travel faster than this speed but not more than 120.4 km/h (74.8 mi/h). The maximum speed on surface streets is 104.6 km/h (65 mi/h). The minimum speed on surface streets is 1.6 km/h (1 mi/h).
CORSIM is not a dynamic traffic model where turn decisions are made based on current conditions. A vehicle will wait in a queue for the duration of the simulation rather than change turning movements to take an alternate path.
TSIS and CORSIM are not geometric design tools. The drawing capability, radius of curve data, superelevation data, and pavement type data are only intended to supply CORSIM with input data for rough geometric estimates for vehicle movement purposes.
Traffic Control
CORSIM’s pre-timed control is limited to 12 intervals, which can be configured to model four sets of green, yellow, and red times. This may be a limitation for intersections with more than four approaches or where more than four phases are required. However, a pre-timed signal with more than 12 intervals can be replicated by coding it as actuated control with all phases on maximum recall. An actuated controller does not have the 12-interval limitation in CORSIM.
CORSIM’s actuated control model is based on a dual-ring, eight-phase controller, as specified in the NEMA TS 1 standard. The CORSIM model can be configured to emulate the operation of the Model 170 controller and many of its features, but the CORSIM terminology is taken from the NEMA specification. The CORSIM model is limited to two rings and eight phases, where phases 1 through 4 are in ring 1 and phases 5 through 8 are in ring 2. However, fewer than eight phases and single ring operation can be coded. The barrier is fixed between phases 1, 2 (5, 6) and 3, 4 (7, 8).
Facilities Not Modeled
There are some types of facilities that cannot be directly modeled in CORSIM, nor can they be indirectly modeled without substantial software development and interfacing with CORSIM via a run time extension.
- Transit signal priority (changing the signal timing based on approaching transit vehicles).
- Light rail facilities.
- Variable message signs.
Facilities Indirectly Modeled
Some facilities that are not explicitly modeled may be modeled indirectly in CORSIM. It may be necessary to use a different tool other than CORSIM that directly models these types of facilities if their presence in the model is critical. Some discussion of these types of facilities is presented in appendix K.
- Roundabouts.
- U-turns.
- Toll booths.
- Weigh station.
- Draw bridge.
- Two-way left turn lane.