Michigan 2006 Five Percent Report
This report is in response to the Federal requirement that each state describe at least 5 percent of its locations currently exhibiting the most severe highway safety needs, in accordance with Sections 148(c)(1)(D) and 148(g)(3)(A), of Title 23, United States Code. Each state’s report is to include potential remedies to the hazardous locations identified; estimated costs of the remedies; and impediments to implementation of the remedies other than costs. The reports included on this Web site represent a variety of methods utilized and various degrees of road coverage. Therefore, this report cannot be compared with the other reports included on this Web site.
Protection from Discovery and Admission into Evidence—Under 23 U.S.C. 148(g)(4) information collected or compiled for any purpose directly relating to this report shall not be subject to discovery or admitted into evidence in a Federal or State court proceeding or considered for other purposes in any action for damages arising from any occurrence at a location identified or addressed in the reports. |
Additional information, including the specific legislative requirements, can be found in the guidance provided by the Federal Highway Administration,
http://safety.fhwa.dot.gov/safetealu/guides/guide040506.cfm.
Michigan 5% Report Methodology 2006 – per Section 148 (c)(1)(D) of SAFETEA-LU
Introduction
This report methodology is prepared by the Michigan Department of Transportation (MDOT), to report on all public roads in the state. The methodology is based on the guidance offered in the April 10, 2006 FHWA 5% Reporting Guidance Document and the subsequent HSIP related web conference held June 20, 2006.
In Michigan, there are approximately 122,000 miles of public roads, of which 92 percent are under local jurisdiction. MDOT does not have any current methodology to conduct statewide surveillance of the road network that includes local roads. As MDOT does not have jurisdiction over local roads, it has not been our practice to develop a statewide list of “hazardous” locations due to the availability of resources to address locations of concern, and the competing priorities of local units of government.
MDOT is in the process of developing new system surveillance tools which will allow review of the entire statewide network. For the purposes of this report a method was developed to make best use of the available information with the objective to identify intersections and segments with the “most severe safety needs”.
In Michigan a statewide GIS referencing system is utilized on which all crashes, on all roads, are referenced to. We have good crash data, located such that we can conduct a statewide analysis to determine the most “hazardous” locations. Michigan crash data contains information on the severity of the crash, whether it is a fatal injury (K), incapacitating injury (A), non-incapacitating injury (B), possible injury (C) or property damage (O) crash. We have information on the number of injuries by severity, if any, suffered by all parties involved in a crash.
MDOT does not have accurate traffic volume data for all roads statewide, thus other exposure measures needed to be considered.
Methodology
Crashes were divided into intersection related crashes and segment crashes. Intersection crashes are identified as a focus area in the Michigan Strategic Highway Safety Plan (SHSP). For future reporting, segment crashes could be broken down into subcategories such as lane departure, another focus in the Michigan SHSP. The Michigan SHSP also identifies pedestrian and bicycle as a focus area. The method used for this years report, identified pedestrian crashes as a concern at a number of locations.
As this is the first year that MDOT has undertaken a statewide study, the methodology that has been developed is fair, given the time and data constraints. MDOT intends to improve the methodology and the report in subsequent years. This will be accomplished by researching other methods, improvement of the local agency contact piece and extending data collection/ integration.
Our preference would have been to develop an “index” or some type of ranking tool that would include at least three factors: crash frequency, crash severity, and crash rate based on vehicle miles traveled (VMT). As noted, we do not have the luxury of statewide VMT data, by route, so MDOT has elected to proceed with the available data. The locations that were identified in this method resulted in analysis of data from crash reports only.
For this inaugural report year, Michigan decided to use K and A incidents to measure safety needs. Rather than have a measure of exposure, we are using a density of crashes per intersection or per one mile segment. Our approach is to develop a combined score for a location (intersection or segment) based on the frequency of K and A crashes and the economic “loss” of K’s and A’s at a location.
The methodology to identify roadway segments with safety concerns was constrained to examining locations where crashes were reported and located. Unlocated crashes are not included in this analysis (currently in Michigan this is a relatively small number).
The method presented discussed here is not unique or original. It is a modified version of location ranking techniques in papers produced by the Iowa DOT (IDOT)1,2.
Locations: Intersections
For intersections, statewide crash data for the years 2003-2005 were used. A crash file containing crashes indicated as “intersection related” was created.
The score for an intersection was determined as follows:
The total number of K and A crashes at an intersection was established. Every intersection statewide with at least one reported K or A crash was included. Locations were ranked, in descending order (the worst ranked 1), by the total number of K and A crashes at the location.
To establish a measure for “loss” the number of fatalities and/or A injuries were used. A straight weighting scheme was used where an assigned dollar value for loss for K’s and a value for A’s was used (values from the National Safety Council website). The number of K’s was in turn multiplied by that dollar loss and the number of A’s by its respective loss value. The two computed values were added.
In an attempt to reduce the impact of the randomness of fatalities we elected to modify the calculation of loss by eliminating the locations where one and only one fatality occurred. In this case, the fatality was counted as an A injury1,2. The value for the variable loss was then ranked with the greatest loss being ranked 1.
Each time a variable was ranked, those locations having the same number of crashes (or alternately the same value for “loss”) were all assigned the low ranking value. For example, intersection locations 3 through 6 had seven crashes so each location was ranked 3. The next rank assigned was 7.
A score was assigned to each location and was the sum of the rank for frequency plus the rank for loss. The lower the value of score the worse the location.
Examination of a histogram of the variable score showed that score was not “normally” distributed. Taking the natural log of the variable score allows one to determine the geometric mean, geometric standard deviation and confidence intervals from a “log normal” distribution.
Taking the 99% confidence interval (the mean minus 3 standard deviations) gives the 295 worst locations in the state. 5% of 295 gives 15 locations, however locations 14 through 17 have the same score so the top 17 locations were used. A 95% confidence interval (the mean minus 2 standard deviations) could have been used, however it was determined that the list of locations would become unmanageable given the reporting time frame. As the crashes for each location were analyzed there were several crashes that were mislocated and the score for such locations were revised. Due to this, a location should have dropped out of the analysis. Rather than drop the location, any locations with the same revised score were added to the analysis – in this case an 18th location was added.
Locations: Segments
2001 – 2005 statewide crash data was used for the segment analysis for all roads.
A study file consisting of crashes indicated as “mid-block” was created for this portion of the analysis. The approach was to accumulate crashes by reference segment number to determine the number of crashes per mile. A “Score” was determined and analyzed for each segment, as described above for intersections.
Taking the 99% confidence interval (the mean minus 3 standard deviations) gives the 212 worst locations in the state. 5% of 212 gives 11 locations, however locations 11 and 12 have the same score so the top 12 locations were used. As the crashes for each location were analyzed there were several crashes that were mislocated and the score for such locations were revised. Due to this, a location should have dropped out of the analysis. Rather than drop the location, any locations with the same revised score were added to the analysis – in this case a 13th location was added.
Following the creation of the list of locations of interest cursory field reviews were conducted. A number of issues arose in the course of the field review portion of the study. This discussion is presented in the “Issues” section of this report.
It should be noted that presently we do not have a good means to approach the segment analysis. The establishment of segment lengths is somewhat arbitrary and is an artifact of our referencing system. We are looking at ways to use some type of sliding or floating interval to isolate concentrations of crashes.
Michigan Highway Intersections / Segments Exhibiting the Most Severe Safety Needs
| Location | Jurisdiction | Potential Remedies | Estimated Costs | Implementation Impediments | Comments | Crash and Injury Data | Human Factors Involvement | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Crash Pattern | Total Severe Crashes | A Injuries | Fatalities | Alcohol and/ or Drugs | NO Seatbelt and/ or Helmet Use* | Speed** | ||||||
| French Rd & Lynch Rd | City of Detroit | Provide signing. Provide a flashing beacon. | $150,000 | The City of Detroit has mitigated the immediate safety concerns by installing guard rails, chevron arrows and removing the huge pile of dirt that was causing the high speed vehicles to become airborn. | Fixed Object | 6 | 3 | 3 | 1 | 0 | 6 | |
| Chicago Dr (M-21) & Cottonwood Dr | MDOT | Remove the flash schedule from the signal operation. Re-evaluate signal operations after interchange project is completed. | N/A | N/A | The Flash Schedule has been requested to be removed. Full actuation may be considered if there is an increase in red-light violation or complaints about unnecessary delays. The Baldwin Interchange Construction will occur in 2007-2008. The new interchange will change traffic patterns and the intersection’s operation will be re-evaluated after the completion of the project, where operational changes will be made at that time. | Angle | 6 | 6 | 2 | 2 | 0 | 0 |
| W Davison St (M-8) & Linwood St | MDOT | Updated signal clearance intervals | There are no practical geometric improvements that can be made at the intersection. There is a Signal Optimization project on Davison in FY 2007 that will update the signal clearance intervals, and help to reduce the crashes. | None | 9 | 11 | 0 | 0 | 3 | 0 | ||
| Southfield Rd & WB 11 Mile Rd | Oakland County | Rebuild this intersection and adjacent intersection on Southfield Road as box spans and use one traffic signal controller | $150,000 for signal improvements | County intends to apply for federal safety funds for this location. | Red Light Running | 7 | 7 | 0 | 1 | 1 | 0 | |
| Highland Rd (M-59) & Bullard Rd | MDOT | Continue median through intersection, increased enforcement, speed reduction measures (pavement markings, landscaping, etc.) | Availability of local law enforcement | Temporary closures may be tested before a permanent closure is considered. Median crossovers are provided approximately 600' on either side of the intersection. | Angle | 4 | 1 | 3 | 0 | 0 | 0 | |
| Jupiter Ave NE & W River Dr NE | Kent County | N/A – Improvements in process | N/A | N/A | The Kent County Road Commission is in process of upgrading the signal to a box span, updating clearance intervals, and revising phasing at this location. This project is currently under construction with a planned completion at the end of the 2006. | Red Light Running | 7 | 9 | 0 | 0 | 1 | 0 |
| S Dixie Hwy (M-125) & Sterns Rd | MDOT | Trim trees to improve sight distance, Investigate signal timing to run in stop/go mode at all times and/or upgrade signal. | $250 for tree trimming. $75,000 for signal upgrade. | Not previously identified | Majority of crashes occurred while signal in flash mode. Signal placed as a temporary signal that now runs in stop/go mode with 10PM to 6AM flash. In 2007 the signal will be retimed as a county-wide retiming project. | Angle | 4 | 4 | 3 | 2 | 1 | 1 |
| Groesbeck Hwy (M-97) &15 Mile Rd | MDOT | Access Management Program | $150,000.00 | A likely cause of the traffic accidents is a high traffic volume combined with the businesses and driveways in close proximity to the intersection. Although not feasible, access management could help to alleviate the accident patterns at this location. | Rear End | 6 | 9 | 0 | 1 | 0 | 0 | |
| Greenfield Rd & Schoolcraft St | Wayne County | Revise signal operations to remove the permitted portion of the left turn phase and provide protected left turns only. | Longer queues for left turning vehicles could potentially create additional conflict points and additional crashes. | Pedestrian and Fail to Yield | 6 | 9 | 0 | 0 | 0 | 0 | ||
| 7 Mile Rd & Beech Daly Rd | Wayne County | Enforcement | Not provided | Availability of local law enforcement | Pedestrian and Bicycle | 4 | 3 | 2 | 3 | 0 | 0 | |
| N Telegraph Rd (US-24) & W Hurd Rd | MDOT | Improve sight distance through shoulder widening/radius improvements. Place advance intersection warning signs or updated W2-1 with 45 mph speed advisory. | $275,000 for shoulder widening. $500 for signing. | Not previously identified | A flashing beacon study has been requested. | Angle | 4 | 2 | 2 | 0 | 1 | 0 |
| S State Rd & E Michigan Ave (US-12) | MDOT | Evaluate signal for left turn phase, box span signal display, signal backplates | $100,000 for signal work | Not previously identified | The signal is to be upgraded to a box span and additional lanes provided in 2007 due to proposed development. | None | 3 | 2 | 1 | 0 | 0 | 0 |
| Ann Arbor Trail & Middlebelt Rd | Wayne County | Box span, evaluate for left turn phase | Not provided | Not provided | The Ann Arbor Trail approaches are to be widened to provide east/westbound center left turn lanes and signal upgrades are planned. Left turn crashes at this location have occurred on the north/southbound approaches. | Head on-Left turn | 4 | 7 | 0 | 0 | 2 | 0 |
| Silver Lake Rd & Division (US-31) | MDOT | Evaluate signal operations to meet seasonal demands. | Red Light Running and Rear End | 5 | 9 | 0 | 0 | 3 | 0 | |||
| Belding Rd (M-44) NE & Lincoln Lake Ave NE | MDOT | Supplement the front and back of the existing stop signs on the Lincoln Lake approaches with “Cross Traffic Does Not Stop” signs (W4-4p). Analyze signal warrants for intersection. | $700 for signing. $75,000 for signal work. | Meeting warrants | A full study for traffic signal installation has been ordered. | 5 | 9 | 0 | 2 | 2 | 0 | |
| N US-31 & Riley St | MDOT | Continue to monitor the intersection. | N/A | N/A | The flash schedule was removed in March of 2006. Full actuation may be considered if there is an increase in red-light violation or complaints about unnecessary delays. | 7 | 8 | 0 | 1 | 0 | 0 | |
| Dexter Ave & W Davison (M-8) | MDOT | Updated signal clearance intervals | There are no practical geometric improvements that can be made at the intersection. There is a Signal Optimization project in FY 2007 on Davison that will update the signal clearance intervals, and help to reduce the crashes. | Pedestrian | 7 | 8 | 0 | 1 | 0 | 0 | ||
| Merriman Rd & Van Born Rd | Wayne County | Enforcement, Evaluate for left turn phase | Signal modernization - $65,000. | Availability of local law enforcement | Head on-Left turn | 7 | 7 | 0 | 1 | 0 | 1 | |
| WB I‑96-Local Lanes E. of Hubbell – W. of Greenfield | MDOT | There was a reconstruction project at this location in FY 2005. We reviewed the crashes after the reconstruction and it seems as if the problem has been addressed. We will continue to review the crashes at this location. | Fixed Object and Pedestrian | 13 | 9 | 6 | 6 | 6 | 3 | |||
| Michigan Ave (US‑12) – Holmes Rd to Wiard Rd | MDOT | Street lighting, sidewalk, additional warning signs | $5,000+ per street light, $4 per square foot of 4" concrete sidewalk, $500 for signs | Availability of funds from township – local agencies are responsible for lighting. | Pedestrian crashes were approximately 1,000 feet apart - midblock pedestrian signal not practical. | Pedestrian | 6 | 3 | 4 | 2 | 1 | 0 |
| Gull Rd (M‑43) – 1200’ E&W of Sprinkle Rd | MDOT | Box span signals for better alignment. Signal optimization throughout corridor. Access management. | $130,000 for signal work. $75,000 for access management. | Receiving the cooperation of the local businesses to close and share driveways. | Signal optimization is on-going for this cooridor. The township has requested an access management study. | 8 | 7 | 3 | 1 | 0 | 0 | |
| Dort Hwy (M‑54) – 500’ N of Hemphill to Mitchell St | MDOT | Pedestrian push buttons and non motorized path | $420,000 for non-motorized path | Non-motorized path may not reduce crash types | Pedestrian | 8 | 7 | 3 | 3 | 1 | 0 | |
| WB I‑96 Express Lanes – Schoolcraft to Outer Drive | MDOT | There was a reconstruction project at this location in FY 2005. We reviewed the crashes after the reconstruction and it seems as if the problem has been addressed. We will continue to review the crashes at this location. | Varied | 8 | 7 | 3 | 1 | 0 | 2 | |||
| State Rd (M‑15) – Atherton Rd to Bristol Rd | MDOT | None | N/A | Highly residential cooridor. Crash types not due to roadway feature. | Head On | 5 | 6 | 3 | 1 | 1 | 0 | |
| EB I‑96 – ½ Mile E/W of Zimmer Rd | MDOT | Investigate feasibility of an early detection system when visibility is zero. | N/A | Extreme events such as this not known to have occurred in this area previously. | One fatal and 3 A type crashes happened on same day under extreme fog conditions which caused multi car pileup along this section of roadway. | Rear End | 7 | 5 | 3 | 1 | 1 | 3 |
| Orchard Lake Rd - Bristol Ln to Ravine Dr | Oakland County | Rebuild from Bristol Ln to 14 Mile as a 6-lane boulevard. | $22,000,000 | Orchard Lake is to be rebuilt as a boulevard from south of 14 Mile to Maple as a boulevard when funding becomes available. | Driveway Related | 5 | 5 | 3 | 0 | 4 | 1 | |
| Alden Rd (CR 618) – Comfort Rd to Grass River Rd | Antrim County | Improve the alignment of the roadway, additional signs | N/A | Environmental constraints prevent changes in alignment. | Signs have been provided. | Fixed Object | 5 | 5 | 3 | 3 | 1 | 4 |
| Lonyo Rd – Dix Rd to John Kronk | City of Detroit | Provide attenuators. | $100,000 | There is a rail road crossing just south of John Kronk. Impact attenuators are scheduled to be installed in year 2007. | Fixed Object and Train | 5 | 4 | 3 | 1 | 2 | 2 | |
| US‑223 – Treat Hwy to Ogden Hwy | MDOT | Add Center Left Turn Lane | $1,400,000 | N/A | A center left turn lane is scheduled for construction for 2007. | Head On | 5 | 3 | 3 | 0 | 1 | 4 |
| 8 Mile Rd – Grand River Ave to Brentwood St | Oakland County | None | N/A | N/A | No correctable crash pattern at this location. | None | 5 | 3 | 3 | 1 | 0 | 0 |
| Rochester Rd (M-150) – 700’ S of Auburn Rd to Regal Ave | MDOT | Access Management Program | $150,000 | Rochester Road is located in a highly congested traffic area of Rochester Hills with businesses, a shopping complex, and restaurant traffic at each corner of the intersection. Although not feasible, access management, sponsored by the City of Rochester could help to alleviate the accident patterns at this location. | Driveway Related | 16 | 19 | 2 | 0 | 2 | 0 | |
Additional Notes:
*Only those drivers or passengers not wearing a seatbelt or wearing a helmet (motorcycle crashes) were reported. Use for remaining injuries could be listed as yes, unknown or not applicable.
**Speed related crashes involve vehicles whose hazardous action was listed at "Speed too fast".
Issues
Our approach to this exercise was to not constrain this analysis any more than necessary. For a number of the locations identified, no correctable crash pattern was identified. Our approach was designed to identify locations that have poor safety performance, not necessarily those with poor safety performance that can be fixed in terms of engineering fixes, or behavioral treatments. A future approach may include focusing on finding locations with correctable crash patterns such as lane departure crashes.
As this is a statewide review, we are also reporting on locations that are under local jurisdictional control. While we (MDOT) have discussed potential fixes, in most cases we have not had the opportunity to discuss treatments, treatment costs, and impediments with local agencies. This effort is currently underway. We have reservations about reporting on local agency priorities, fixes, and cost estimates.
At some of the locations identified in our methodology the safety treatments have already been applied. In our future reports, additional strategies to filter out these locations, prior to the field review will need to be investigated. Future reporting will involve MDOT personnel closer to the locations in question to take advantage of their local knowledge of conditions.
In many cases a potential treatment at a signalized location is to install a box span signal display. It is now standard in Michigan to install box spans at trunkline intersections where the signal span is to be upgraded. Also, on-going signal optimization projects on all trunklines (to be completed within five years) may mitigate some of the crashes at signalized locations.
Michigan’s crash data base system is in good condition. A new system completed in 2004 has a number of features which will allow for the continuous improvement of data quality and accuracy.
The next step for Michigan is to better integrate existing traffic safety databases, and extend the coverage to the entire network. Discussions are under way to begin the planning for this action. Of primary importance will be the development of plans to collect, or access ADT/VMT data. While many local agencies collect traffic count information in some form, there are no standards in place for collecting and adjusting count information.
We will continue to improve our method of determining our 5%. More work is required on the fundamental approach, and conducting sensitivity analysis. Our first application of this methodology suggests that it is sensitive, but did consistently yield locations with safety problems.
Contact person
Dale Reed Lighthizer, P.E.
MDOT – Traffic and Safety
517.373.2334
lighthizerd@michigan.gov
References
1Iowa State University CTRE. Systematic Identification of High Crash Locations. Iowa Department of Transportation. May 2001.
2Iowa State University CTRE. Evaluation of the Iowa DOT’s Safety Improvement Candidate List Process. Iowa Department of Transportation. June 2002.
