Analysis, Modeling, and Simulation for Traffic Incident Management Applications

Incident Data Required to Support TIM AMS Applications

TIM Performance Measures

Monitoring trends in performance leads to sound investment in transportation strategies and policies. Performance programs have evolved over the years to include many types of performance measures that are used to manage incident programs. Recently, there has been a move to developing a few core measures that can be used as indicators of program performance for all transportation agencies, while recognizing that additional measures are needed at the agency level. In TIM, a national effort has produced three of these core measures:

• Roadway clearance time is the time between first recordable awareness of an incident (detection/notification/verification) by a responsible agency and first confirmation that all lanes are available for traffic flow.
• Incident clearance time is the time between the first recordable awareness and the time at which the last responder has left the scene.
• Secondary crashes are those that occur with the time of detection of the primary incident where a collision occurs either a) within the incident scene or b) within the queue, including the opposite direction, resulting from the original incident.

Additional TIM performance measures are of use to agencies beyond the core measures. Monitoring other parts of the incident “timeline” also can lead to identifying and correcting program deficiencies. Incident locations are used to identify hotspots. Identifying types of incidents (e.g., crashes, vehicle disablement, hazardous material release, large truck involvement) also produces actionable information.

Data Collection

The incident data to support the development of performance measures are the same that are required to support TIM modeling applications and includes the types of data shown in Figure 5.

Incident Characteristics – This data type includes elements that describe the number, location, characteristics, and duration of incidents. The most important data elements for the research are:

• Incident type (crash, disabled vehicle, fire, debris, abandoned vehicle).
• Incident collision type (fixed object, overturn, vehicle/side, vehicle/head-on, vehicle/rear-end).
• Incident “timeline” data (time stamps for: start of incident, detection, verification, on-scene arrival, lane/shoulder open, all clear).
• Incident severity (KABCO injury scale).
• Incident location (route, travel direction, milepost or GPS coordinates).
• Incident blockage. (Ideally, blockage is updated every time blockage changes during a given incident.)
  • Cross section feature affected (lane, partial lane, right shoulder, left shoulder, median, off maintained way);
  • Lane type (through/GP, through/HOV, auxiliary, on-ramp, off-ramp); and
  • Begin/end time.
• Number of involved vehicles.

Figure 5. Incident Data Model for Performance Monitoring

(Source: Cambridge Systematics, Inc.)

The duration of the blockages caused by incidents is the most important piece of information needed for research, but others also affect traffic flow conditions. Also, some incident data may be incomplete and the type of incident may be used as an indicator of the lane closure impacts (e.g., for data checking purposes).

Possible sources of data include the following:

• Service Patrol/Incident Response Team Data Entry – PDAs/laptops in the vehicles;
• Freeway service patrols (on-scene reporting);
• Law enforcement officer on-scene reports;
• TMC Operator Data Entry – Automate the conversion of web information to database entries; and
• Police Computer-Aided Dispatch (CAD) files.

Incident Activity Data – This data type relates to the nature of how incidents are managed by an agency. Having this type of data is not strictly necessary for doing incident modeling, as they are rarely used as direct input variables to modeling procedures. However, incident activity data can influence incident characteristics. For example, increasing the number of service patrols can reduce incident response and on-scene management times. Many past studies have established links between incident activities and incident characteristics; the Joint Program Office’s ITS benefits database has a compilation of these relationships.

Examples of the incident activity data include:

• Incident detection algorithms (used with traffic sensors);
• Method of incident notification to all responders (communication between agencies);
• CCTV coverage (percent of study segment capable of being viewed with CCTV);
• Degree of wireless communication for on-scene management;
• Service patrol vehicles per centerline mile by time period;
• Quick clearance laws or policies;
• Length of time abandoned vehicles are allowed to remain on a freeway shoulder (assuming they are not an imminent hazard);
• Laws or policies regarding the removal of stalled or abandoned vehicles from freeway shoulders;
• Policies and procedures to facilitate quick removal of heavily damaged vehicles and nonhazardous cargoes; and
• Procedures for removing crash victims from incident scene.

Other data inputs to models and analytic procedures – Table 3 lists some of the other data inputs needed to run the various scales of AMS procedures.

Table 3. Input Requirements of TIM AMS Procedures

Model Type	Network	Demands	TIM	Calibration
Sketch Planning Models	Minimal. Generally deals with regional VMT and VHT.	Minimal. Generally regional VMT.	General categories of TIM strategies. No specific implementation details.	Not applicable.
Deterministic (HCM Type Macroscopic) Models	Link and intersection-specific lane geometry, speed limits, controls (signal timing).	Link and intersection-specific hourly demands by vehicle type (usually just peak hour).	Incident types (number of lanes blocked, specific links affected, and average duration). Strategies to be tested and expected effects on average lane blockage durations.	Not generally done.
Mesoscopic Simulation Models	Same as deterministic HCM.	OD tables by hour of day for peak periods.	Same as for deterministic HCM.	Observed flows and link speeds.
Microscopic Simulation Models	Same as HCM plus signal detector locations, signal controller settings, turn pocket lengths.	Same as mesoscopic or link and intersection-specific demands by vehicle type.	Incident start/end times, longitudinal location within link. Expected effect of strategies on specific incident duration.	Observed flows and link speeds.

Default Data for TIM AMS Applications

In the absence of agency-specific data, data developed from other areas may be used for TIM AMS applications, especially for forecasting purposes. Some of these data represented below. SHRP 2 Project L08 is developing a larger library of these data.

A major caveat in the use of these data is that they were developed from areas where active incident management programs already exist. Thus, the incident duration data in particular may not be representative of a “before” case where active TIM does not exist.

For planning applications, it is often necessary to predict the number of future incidents. This can be done in a number of ways, including the following methods:

• Use of Incident Rates. If the analyst has the current incident rates (incidents per vehicle-miles of travel, VMT), either for a specific facility or as an areawide default, the number of incidents can be estimated simply by multiplying the incident rate by the forecasted VMT.
• If overall incident rates are not available, but a crash rate is, the crash rate can be factored up to total incident rates using agency-developed default values. If these default values are unavailable, the national default factors being developed by SHRP 2 Project L08 can be used. In the interim, a factor of 5.0 can be used, based on analysis of Atlanta data that showed crashes are roughly 20 percent of all incidents.
• If stall rates are not available, the analyst should use the procedures in the Highway Safety Manual to estimate the number of crashes, convert this to a crash rate, then factor up to a total incident rate.

Table 4. Lateral Locations of Incidents, Atlanta, Georgia

Incident Type	1 Lane	2 Lanes	3+ Lanes	Off Roadway	Shoulder	Total
Accident	Frequency: 2,566 Row Percent: 44.07 Column Percent: 28.32	Frequency: 1,451 Row Percent: 24.92 Column Percent: 69.19	Frequency: 1,052 Row Percent: 18.07 Column Percent: 84.77	Frequency: 41 Row Percent: 0.70 Column Percent: 21.47	Frequency: 712 Row Percent: 12.23 Column Percent: 4.39	Frequency: 5,822
Stall	Frequency: 4,411 Row Percent: 21.81 Column Percent: 48.68	Frequency: 261 Row Percent: 1.29 Column Percent: 12.45	Frequency: 100 Row Percent: 0.49 Column Percent: 8.06	Frequency: 101 Row Percent: 0.50 Column Percent: 52.88	Frequency: 15,349 Row Percent: 75.90 Column Percent: 94.72	Frequency: 20,222
Debris	Frequency: 2,047 Row Percent: 77.30 Column Percent: 22.59	Frequency: 382 Row Percent: 14.43 Column Percent: 18.22	Frequency: 87 Row Percent: 3.29 Column Percent: 7.01	Frequency: 47 Row Percent: 1.77 Column Percent: 24.61	Frequency: 85 Row Percent: 3.21 Column Percent: 0.52	Frequency: 2,648
Roadkill	Frequency: 37 Row Percent: 35.92 Column Percent: 0.41	Frequency: 3 Row Percent: 2.91 Column Percent: 0.14	Frequency: 2 Row Percent: 1.94 Column Percent: 0.1	Frequency: 2 Row Percent: 1.94 Column Percent: 1.05	Frequency: 59 Row Percent: 57.28 Column Percent: 0.36	Frequency: 103
Total	Frequency: 9,061	Frequency: 2,097	Frequency: 1,241	Frequency: 191	Frequency: 16,205	Frequency: 28,795

Table 5. Incident Duration by Incident Type, Atlanta, Georgia

	Lateral Location	Frequency	Duration (Minutes)
			Mean	Standard Deviation	Median	95th Percentile
Accident	1 Lane	2,566	38.942	28.526	34.700	89.367
	2 Lanes	1,451	46.279	30.697	41.283	96.450
	3+ Lanes	1,052	66.718	56.219	56.233	149.767
	Off Roadway	41	14.377	20.580	4.267	71.817
	Shoulder	712	40.112	34.306	34.375	91.800
	TOTAL	5,822	46.287	38.129	40.033	106.067
Stall	1 Lane	4,411	29.055	31.382	24.100	72.083
	2 Lanes	261	40.650	44.884	33.917	109.883
	3+ Lanes	100	40.002	23.263	38.850	76.800
	Off Roadway	101	30.753	28.419	25.050	97.533
	Shoulder	15,349	41.506	29.356	35.483	96.300
	TOTAL	20,222	38.996	30.337	32.917	93.350
Debris	1 Lane	2,047	16.588	21.955	11.408	49.000
	2 Lanes	382	17.789	23.002	11.833	51.067
	3+ Lanes	87	45.448	62.059	22.833	183.633
	Off Roadway	47	5.027	4.091	2.717	10.567
	Shoulder	85	35.713	46.699	19.958	175.883
	TOTAL	2,648	18.797	27.375	11.833	61.433
Roadkill	1 Lane	37	47.064	57.934	32.650	226.883
	2 Lanes	3
	3+ Lanes	2	2.325	1.473	2.325	3.367
	Off Roadway	2	20.650		20.650	20.650
	Shoulder	59	33.922	63.202	8.575	121.850
	TOTAL	103	36.292	60.359	13.858	135.317