BACKGROUND 

Statistics from the Fatality Analysis Reporting System (FARS) indicate that more than 25 percent of fatal crashes occur at horizontal curves, with most of these crashes being roadway departures. Although researchers and practitioners agree that curvature plays a role in crash frequency and severity, safety performance functions (SPFs) and severity distribution functions for horizontal curves have not been thoroughly investigated or widely implemented. 

The American Association of State Highway and Transportation Officials (AASHTO) Highway Safety Manual (HSM) provides SPFs of various facility types for segments and intersections, but not for curve segments. Rather, curves are evaluated by applying an adjustment factor (AF) to estimate the predicted crash frequency of a curve segment. Unfortunately, not all the SPF models within the HSM have AFs for horizontal curves. Recent studies have been implemented to develop AFs for curve segments of certain facility types to begin filling this gap. However, applying a horizontal curve AF to an existing segment SPF assumes that the underlying prediction model of a tangent segment only needs to be adjusted to appropriately estimate a horizontal curve’s influence on the segment’s safety performance. This method may not be the best way to assess the safety performance of horizontal curves. A more thorough investigation may reveal that common geometric attributes used to estimate the safety performance of tangent segments have a different degree of influence on the safety performance of horizontal curves. Taking this possibility further, the attributes most important for predicting the safety performance of a horizontal curve may differ due to context (rural versus urban), facility type, or other factors such as geometric and operational characteristics, and roadside elements. Therefore, additional research is needed to better understand the attributes that most influence the safety performance of horizontal curves.

OBJECTIVE 

The objective of this research is to develop a guide and a tool to quantify the safety performance of horizontal curves using geometric and operational characteristics for applications across a range of highway activities including planning, design, operations, and safety management

At the minimum, the research shall: 

1. Advance the predictive safety performance methodologies for horizontal curves on rural and urban freeways, rural and urban multilane divided roadways, rural and urban multilane undivided roadways, and rural and urban two-lane undivided roadways. 
2. Develop and validate a statistically valid predictive methodology to quantify the safety performance of locations with curves for use by state and local transportation agencies to help evaluate the likely safety performance of different horizontal curves at a given location. The study should consider geometric and operational characteristics of contiguous highway segments and various facility types. Consideration should be given to different horizontal curve types such as simple curves, compound curves, reverse curves, and broken-back curves. 

 Accomplishment of the project objective will require completion of the following tasks, at a minimum.

TASKS 

PHASE I—Planning 

Task 1. Conduct a critical review of the road safety literature to identify factors associated with the safety performance of horizontal curves. The review should summarize the factors considered, discuss related modeling efforts, and identify gaps in the current state of knowledge and practice on the research topic. 

Task 2. Identify data currently used as well as other public data that could help develop exposure and safety prediction models for use in locations with horizontal curves. Develop a comprehensive data collection, management, and validation plan. The plan should include metadata for raw and processed data, data ownership information, data restrictions (if applicable), and recommendations for data archiving.

Task 3. Prepare a detailed Phase II work plan that describes a methodological framework for developing safety prediction models. Examine key factors for evaluating the safety performance of horizontal curves, using geometric and operational characteristics.

For the modeling task, the proposed Phase II work plan should include, at a minimum:

• Methods for multi-source data integration/fusion;
• Data collection, management, and distribution plan;
• Identified tools or software to be used for data management, processing, and modeling;
• Metrics for assessing statistical models and analysis results;
• Validation approach for the methodology and tool; and
• Development of a spreadsheet tool.

The methods for prediction of crash frequency and severity for various facility types should be repeatable, be capable of evaluating existing and future conditions, and be practical and readily implementable by state and local transportation agencies of all sizes to help evaluate potential safety performance at a given location.

Task 4. Prepare Interim Report No. 1 documenting the results of Tasks 1 through 3 and provide an updated plan for the remainder of the research no later than 6 months after contract award. The updated plan must describe the methodology and rationale for the work proposed for Phases II and III. 

PHASE II—Study 

Task 5. Execute the work plan based on the approved Interim Report No. 1. 

Discuss in detail the impact of the significant variables, including sign of impact (+/-), magnitude, and elasticities (or marginal effects), with the purpose of improving the scientific understanding of the safety performance of locations with horizontal curves. Include in the documentation for model development:

• Detailed methodology;
• Data summary and descriptive statistics by facility type and horizontal curve geometric characteristics;
• Validated models and results, including variable importance ranking and the elasticities (or marginal effects) of analyses by facility type;
• Metrics used for assessing the quality of statistical models and the results of quantitative analyses; and
• Tables and graphs showing relationships between crash frequency/severity and key independent variables.

Provide all assumptions, data limitations, and other constraints (e.g., the range for valid input fields, conditions that could not be assessed with the methodologies because the data sources used to develop the method did not include these conditions).

Task 6. Develop a user-friendly, updatable, and easy-to-maintain spreadsheet tool and user manual that can be used and maintained by state and local transportation agencies for practitioners to quantify the safety performance of horizontal curves using geometric and operational characteristics for applications across a range of highway activities including planning, design, operations, and safety management. The spreadsheet tool should not be developed with third-party software, but with commonly available software.

Task 7. Prepare an annotated outline of the draft guide.

Task 8. Prepare Interim Report No. 2 that documents Tasks 5 through 7 and provides an updated work plan for the remainder of the research no later than 12 months after approval of Phase I. The updated plan must describe the process and rationale for the work proposed for Phase III. 

PHASE III—Reporting

Task 9. Prepare and submit the draft guide based on the approved Interim Report No. 2 for the NCHRP panel’s review.

Task 10. Refine and update the draft guide and the spreadsheet tool based on comments from the NCHRP project panel. 

Task 11. Prepare draft language for consideration by AASHTO to incorporate the research results in the next update of the AASHTO HSM (herein called AASHTO Deliverable). Include sample problems, assumptions, effect sizes, data limitations, and other constraints (e.g., the range for valid input fields, conditions that cannot be assessed with the methodologies because the data sources used to develop the method did not include these conditions). 

Task 12. Present the research findings to appropriate AASHTO technical committees for comments and propose any revisions to NCHRP. The research team should anticipate making two presentations during the research to appropriate technical committees at annual meetings of the AASHTO Committee on Safety or Committee on Design. Revise the draft research report after consideration of review comments.  

Task 13. Prepare the final deliverables including the following:

1. A conduct of research that documents the entire research effort and any lessons learned; 
2. The final guide;
3. The AASHTO Deliverable;
4. All raw and cleaned data collected and used in this research (data should be provided in as close to its raw form as possible, based on contractual or legal restrictions): input data sets, fused and integrated research data sets, data dictionaries, data models, etc;
5. Documentation of the steps in the data fusion and integration process, including any annotated data management codes used for fusing, integrating, and cleaning data;
6. Annotated model development and validation code; 
7. The spreadsheet tool (including any annotated spreadsheet macro code, if used), user manual, and any other spreadsheet tool documentation;
8. Media and communication material (e.g., presentations, 2-page executive level flyer, graphics, graphic interchange format, press releases); and
9. A stand-alone technical memorandum titled “Implementation of Research Findings and Products”. Additional funding may be available for a follow-up contract on the implementation of the results.

STATUS: Research in progress.