[Previous: Crash Prediction Module Engineer's Manual] | [Table Of Contents] | [Next: Data Input Requirements] |
This engineer's manual is one component of the documentation supporting the Interactive Highway Safety Design Model (IHSDM). This introductory section: (1) provides a brief overview of IHSDM, (2) summarizes the capabilities and intended uses of the Crash Prediction Module (CPM), and (3) states the purpose and organization of this engineer's manual.
IHSDM is a suite of software analysis tools for evaluating safety and operational effects of geometric design in the highway project development process. The scope of the current release of IHSDM is two-lane rural highways.
IHSDM is intended as a supplementary tool to augment the design process. This tool is designed and intended to predict the functionality of proposed or existing designs by applying chosen design guidelines and generalized data to predict performance of the design. This tool is NOT a substitute for engineering judgment and does not create a standard, guideline or prescriptive requirement that can be argued to create any standard of care upon a designer, highway agency or other governmental body or employee. The use of this tool for any purpose other than to aid a qualified design engineer in the review of a set of plans is beyond the designed scope of this tool and is not endorsed by the Federal Highway Administration (FHWA).
The suite of IHSDM tools includes the following evaluation modules. Each module of IHSDM evaluates an existing or proposed geometric design from a different perspective and estimates measures describing one aspect of the expected safety and operational performance of the design.
Intended users of IHSDM results are geometric design decision makers in the highway design process, including project managers, planners, designers, and reviewers. The Federal Highway Administration's Flexibility in Highway Design document (Publication No. FHWA-PD-97-062) explains the context within which these decision makers operate:
The measures of expected safety and operational performance estimated by IHSDM are intended as inputs to the decision making process. The value added by IHSDM is in providing quantitative estimates of effects that previously could be considered only in more general, qualitative terms. The advantage of these quantitative estimates is that, when used appropriately by knowledgeable decision makers, they permit more informed decision-making.
The following general cautions should be considered in using IHSDM:
The CPM estimates the frequency and severity of crashes that would be expected on a highway considering its geometric design and traffic characteristics. The crash prediction algorithm combines base models and accident modification factors (AMFs). Jurisdictions may enter calibration factors to adjust estimates to be comparable to the their reported crash experience. The algorithm also provides a procedure to combine model estimates with highway-specific crash data.
Separate base models have been calibrated, using state-of-the-art statistical techniques, for highway segments and three types of at-grade intersections: three-leg intersections with stop control on the minor-road approach, four-leg intersections with stop control on the minor-road approach, and four-leg signalized intersections. The base models estimate the total crash frequency during a specified time period on a highway segment or at an intersection under base geometric conditions and actual traffic conditions.
The AMFs adjust the base model estimates for individual geometric design dimensions and traffic control features. The factors are the product of expert panels, which reviewed related research findings, and reflect their consensus estimate of the quantitative safety effects of each design and traffic control feature.
Each base model was developed using data from one or two States. Reported crash frequencies for nominally similar highway segments or intersections are known to vary widely among States due to differences in climate, animal population, driver populations, and accident reporting thresholds and practices. Therefore, each State should develop and input their own calibration factor, in order to adjust model estimates to be more comparable to their crash experience.
The final component of the Crash Prediction Module is an Empirical Bayes procedure to combine model estimates with locally available crash data for an existing highway. Locally available crash data reflect highway-specific variables that are not accounted for by the base models or accident modification factors. In theory, a weighted combination of model estimates and locally available crash data should produce a more accurate estimate of expected crash frequency than either one by itself.
The module provides a quantitative basis for comparing the expected safety performance of design alternatives. Intended uses include comparison of project-wide crash estimates against State-wide crash experience, comparisons of the relative safety performance of individuals segments or intersections within an existing or proposed design that may merit more detailed evaluation, and safety cost-effectiveness evaluations of the costs versus the safety benefits of design alternatives.
Since, in the past, expected crash frequencies have not been routinely available to project decision makers, there is a need to use caution and gain experience with this measure of safety performance. Some general cautions that should be exercised include:
This Crash Prediction Module (CPM) Engineer's Manual documents the basic information that users should understand in order to make appropriate use of the Module. It details the data input requirements, explains the procedural elements of the module, enumerates the steps in the crash prediction algorithm, and describes the presentation of model outputs. The manual highlights limitations of the Module that users should consider in applying it and interpreting results.
The manual is organized as follows:
[Previous: Crash Prediction Module Engineer's Manual] | [Top] | [Next: Data Input Requirements] |