The TII268 Lot 1 Collision Prediction Model for the Irish National Road Network was carried out by Transport Infrastructure Ireland (TII) as part of its Road Safety Programme. The project ran in two main phases, with the first phase report published in 2022 and the final Phase 2 report released in 2025. TII led the work, coordinating data acquisition, statistical modelling and the development of a user‑friendly tool for practitioners. The project was funded by the Irish government through the national transport infrastructure budget.
In Phase 1 the team assessed the quality of existing crash and road‑feature data and explored suitable statistical techniques. The low traffic volumes and sparse crash counts typical of the Irish network presented challenges, but the analysis confirmed that regression‑based Accident Predictive Models (APMs) could still provide useful safety estimates. Phase 2 involved segmenting the network into homogeneous sections defined by flow, curvature and lane count. Thresholds were applied to reduce the number of zero‑crash sections, and all crash severities were included to maximise data for modelling.
A range of models was tested, and a zero‑inflated negative binomial approach delivered the best fit, particularly for motorways where crash density was highest. The final models were produced separately for motorways, dual carriageways, single carriageways and legacy roads. For example, the motorway network comprised 371 segments, of which 25 (7 %) had no crashes; the mean crash count per segment was 14.7. Dual carriageways had 190 segments with 10 (5 %) zero crashes and a mean of 23.8 crashes per segment. Single carriageways and legacy roads had higher proportions of zero‑crash sections (15 % and 19 % respectively). Key predictors across all road types were geometry (gradient and radius), skid resistance (CSC % values), and the density of minor junctions and access points. The models showed that increasing median barrier coverage on dual carriageways and improving skid resistance on single and legacy roads could reduce collision risk. The findings also highlighted the importance of addressing minor junctions and access points to lower overall crash rates.
The statistical outputs were incorporated into a Collision Reduction Calculator built in Microsoft Excel and made available on the TII website. The tool allows engineers to estimate the expected crash reduction from specific countermeasures, using the Irish‑derived Crash Modification Factors (CMFs). Stakeholder engagement was integral: an online survey and two workshops with TII road safety staff identified user needs such as avoiding double counting of benefits, preference for a First‑Year Rate of Return appraisal, and the inclusion of countermeasures not directly modelled. These insights guided the tool’s design, ensuring it met practical decision‑making requirements.
Future work will focus on enriching the models with additional data, such as traffic speed, user‑type flows, and detailed junction characteristics. The team also plans to refine the tool by providing countermeasure effectiveness broken down by crash sub‑types, allowing users to specify overlap between measures, and linking to external benefit databases. By continuously updating the models and the calculator, TII aims to support transparent, evidence‑based road design upgrades that enhance safety across the Irish network.