The methodology proposed in this paper integrates reactive and proactive approaches to support decision-making for urban road safety improvements. This structured process identifies critical areas (black spots) and evaluates risks, enabling optimal resource allocation and prioritization of interventions. The methodological phases include the following:
While the iRAP star rating methodology is a valuable tool for proactive risk assessment, it has limitations, particularly in complex urban environments. The model is primarily designed for highway and arterial road assessments, which can make it less precise for dense city networks with frequent intersections, varying traffic conditions, and mixed road users. Additionally, iRAP does not account for real-time traffic flow variations or signal timing effects, which are crucial in urban safety assessments. To mitigate these limitations, field observations and complementary GIS analyses were incorporated to refine risk evaluations.
3.2. Proactive Approach to Road Safety
The proactive approach focuses on identifying and mitigating road safety risks before accidents occur. Unlike reactive methods that rely on historical crash data, proactive methods assess the intrinsic safety of roads by analyzing geometric features, traffic flow, and speed. This enables the estimation of potential fatalities, serious injuries, and the associated social costs, supporting the design of targeted interventions to improve safety. For this matter, the iRAP framework has been adopted.
The International Road Assessment Program (iRAP) combines five tools (i.e., crash risk mapping, star ratings, FSI, SRIP, and performance tracking) to enhance road safety globally, particularly for VRUs.
The assessment of road characteristics is based on “coding”, which involves analyzing georeferenced images sampled every 100 m along the road segment. These images, combined with traffic flow data and field measurements, form the foundation for safety evaluations. The steps of iRAP methodology are presented in
Figure 3.
3.3. Black Spots Identification and Risk Quantification
The identification of high-risk areas in a road network can leverage both reactive and proactive approaches. The reactive approach focuses on analyzing historical crash data to objectively assess risk. However, its effectiveness is limited in contexts with insufficient data or in high-risk areas (e.g., dangerous intersections) where caution may lower incidents without eliminating underlying dangers. On the other hand, the proactive approach evaluates intrinsic road safety by analyzing infrastructure, traffic flow, and speed. It is especially valuable where crash data are unavailable or for systematically reducing risks regardless of past incidents.
The Network-Wide Assessment (NWA) guidelines [
26], introduced by the European Commission in 2023, propose combining these approaches to prioritize interventions. The NWA evaluates road risk using a five-level classification system, integrating crash data (when available) with proactive safety assessments. The process involves verifying the availability of at least three years of reliable crash data, then applying both approaches where data exists or using only the proactive method otherwise, and, finally, creating a risk matrix (see
Figure 4) to classify road segments by intervention priority.
The integration of the two approaches ensures that critical points identified by the reactive method are automatically assigned a condition of ‘Very High Priority’, irrespective of the outcomes of proactive assessments. Consequently, the reactive approach was selected as the primary method for identifying high-risk locations within the road network. The proactive approach, in turn, is reserved for subsequent stages, including validation and cost–benefit analysis.
Given sufficient crash data, the reactive approach is recommended for identifying immediate intervention areas. Using GIS tools and heatmaps, high-incident zones are visualized, enabling efficient targeting of problem areas.
The Kernel Density Estimator (KDE) is a statistical method used to create heatmaps that visually represent the density of road incidents across specific areas. By analyzing geospatial data, KDE identifies black spots—areas with a high concentration of accidents [
27]—guiding the prioritization of safety interventions.
To create heatmap, KDE combines individual probability density functions for each data point to generate a smooth visual representation of accident distribution. The bandwidth parameter determines the spread of the density curve, balancing localized detail and broader patterns. Incidents are divided into four categories based on road users involved (i.e., pedestrians, bicycles, motorcycles, and motor vehicles). This disaggregation highlights specific vulnerabilities and tailors interventions to address the needs of each group. Rather than merely evaluating accident rates relative to traffic flow, the analysis prioritizes reducing severe injuries and fatalities. By focusing on the absolute reduction in social costs, interventions target areas with the highest societal impact.
After identifying the black spots through reactive analysis by adopting historical crash data to create density maps (heatmaps) that highlight areas with high accident concentrations, they will be classified as high-risk, requiring immediate attention regardless of subsequent evaluations. At this point, the reactive findings should be validated and intrinsic road safety based on infrastructure characteristics, traffic flow, and speed should be evaluated through iRAP star rating, which is a proactive safety evaluation tool.
During this process segments are scored from one to five stars for safety based on user categories (drivers, motorcyclists, cyclists, pedestrians). The star rating evaluates over 50 attributes grouped into categories such as road edges, carriageways, intersections, traffic flow, vulnerable user facilities, and speed. Factors like lane width, visibility, pedestrian crossings, and traffic density contribute to the risk score. In this study, iRAP’s ViDA platform is adopted to calculate risk scores by using georeferenced images and field data.
The star rating score (
SRS) developed by iRAP quantifies the relative risk of fatal and serious injuries for each road user (drivers, motorcyclists, cyclists, and pedestrians) across different road segments. It is calculated by summing risk scores for various crash types using Equation (1).
These scores represent the risk of specific crash types based on road and traffic conditions. Common crash types include the following:
Motor Vehicles (Run-off-road, head-on collisions, intersection/access-point crashes);
Motorcycles (Run-off-road, head-on collisions, intersection crashes, and roadside incidents);
Cyclists and Pedestrians (Roadside crashes, intersection crashes, and crossing-related incidents).
Each
Crash Type Score is influenced by multiple
Risk Factors, determined by road attributes and calculated using Equation (2).
Risk Factor categories are probability (evaluates the likelihood of an incident based on road attributes); severity (assesses the potential harm caused by an incident); operating speed (considers how speed variations affect risk); traffic flow (accounts for the influence of road usage by other users; and median crossing (examines the likelihood of vehicles crossing the median, particularly in head-on or run-off-road incidents, relevant for drivers and motorcyclists).
The numerical result of SRS is translated into a five-star scale, with each star represented by a distinct color ranging from black (worst) to green (best) (see
Table 1). This color-coding enhances the visual impact and clarity of the ratings, allowing for quick identification of the most critical safety concerns.
When proactive results differ from reactive findings, further investigation is conducted to identify non-infrastructural causes, such as recurring human errors, through field visits. Regardless of proactive analysis outcomes, high-crash areas identified reactively are treated as high-priority for interventions. Reliable inputs are critical for accurate analysis, as incomplete or imprecise data compromises results. The iRAP models may struggle with complex road systems like those in Italy. Adjustments to local contexts are essential for effectiveness. Current methods lack detailed traffic flow analysis and do not account for elements like intersection geometry or traffic light timing, limiting risk assessment precision. The iRAP lacks specific criteria for evaluating cyclist crossings, potentially misrepresenting safety risks. Temporary reliance on pedestrian star ratings addresses this gap but introduces inconsistencies. While iRAP recognizes speed as a key factor in crash severity, its sensitivity to speed variations in low-traffic urban areas is limited. Tools for speed management, such as traffic-calming measures, enhance the model’s utility.
3.4. Cost–Benefit Analysis for Road Safety Interventions
By combining proactive risk evaluation and cost–benefit analysis, it is possible to prioritize road safety measures effectively. By quantifying the social and economic benefits of the proposed interventions, the decision-making process is guided toward solutions that maximize safety while optimizing resource allocation.
3.4.1. Proactive Estimation of Fatalities and Serious Injuries
The proactive cost–benefit analysis estimates reductions in fatalities and serious injuries (FSIs) resulting from road safety interventions. Moreover, it quantifies the associated social cost savings, including human, healthcare, productivity, administrative, and material costs. It considers implementation costs to ensure financial sustainability.
FSIs can be estimated through iRAP’s SRS to predict the annual number of fatalities and serious injuries for four road user categories (drivers, motorcyclists, cyclists, and pedestrians). Calculations incorporate traffic flow (Annual Average Daily Traffic—AADT), crash type risks, and calibration factors tailored to local conditions.
Calibration factors adapt iRAP’s international model to local contexts by integrating historical crash data with proactive risk assessments. Furthermore, it adjusts predictions based on factors like vehicle safety standards and user behavior in specific regions.
One of the novelties of the proposed model is the estimation of serious injury, which relies on a ratio of serious injuries to fatalities (5.5:1 in Italy, based on historical data) to estimate injury figures. In fact, it uses the Maximum Abbreviated Injury Scale (MAIS 3+) to classify serious injuries.
The process described above can be applied to different design solutions, using their respective SRS to assess the associated road safety risks. This approach enables the estimation of the annual number of FSIs after implementing various design alternatives.
For the calculation of the annual number of FSI under the current scenario (Option 0—Do Nothing), the proactive iRAP approach has been chosen. The available Istat dataset provides the number of fatalities for each road segment, but it does not distinguish between minor and serious injuries. Serious injuries are reported only in aggregate for the entire national road network without breakdowns for individual segments. To ensure consistency and comparability of results, the FSI for the “Do Nothing” option is calculated using the same methodology as the other design options, based on risk analysis from the SRS. This ensures that the reactive and proactive approaches remain separate in the cost–benefit analysis, providing a fully proactive calculation independent of historical incident data. The calculation of avoidable FSIs is performed by Equation (3).
where
FSIsaved is the estimate of avoidable FSIs per year;
FSIbefore represents the number of FSIs currently estimated annually; and
FSIafter indicates the estimated number of FSIs per year after the implementation of a road improvement intervention.
Estimating the number of avoidable FSI is a critical step in demonstrating the effectiveness of a proposed road safety solution. However, during the initial decision-making phase, particularly in the technical-economic feasibility study, the choice between different design options must also consider the financial resources available. Therefore, evaluating the benefit of an intervention should go beyond just the number of avoided FSIs and include an economic quantification of the results.
3.4.2. Reduction in Social Costs
This approach provides a monetary value for preserving human lives and reducing serious injuries, offering a useful metric for comparing different solutions and justifying investments. Thus, it is recommended to adopt an approach that centers on the reduction in the social costs arising from road accidents as a key criterion in evaluating design alternatives. Social cost is a complex indicator that translates the financial burdens society faces due to accidents, including healthcare, insurance, legal costs, and lost productivity. Specifically, the loss of productivity refers to an individual’s inability to contribute to the country’s GDP following an accident [
28].
In Italy, the methodology for calculating the social cost of road accidents was defined by the Ministry of Infrastructure and Transport in 2022 [
28] and formalized by the Director’s Decree No. 37 in 2023. To apply these social cost values in the analysis, the most recent data from 2018 [
29] were used, indicating the following reference values.
Social cost per person died in a road accident €1,812,989;
Social cost per person seriously injured €467,159;
Social cost per person lightly injured: €8519.
According to Istat [
4], the total social cost of road accidents in Italy in 2023 was nearly €18 billion, equivalent to 1% of the national GDP. This underscores the importance of implementing effective road safety policies to alleviate the economic burden on society.
It is important to note that the iRAP model used in the proactive analysis does not account for minor injuries, which were excluded from this study. However, although the social cost of minor injuries is much lower than that of serious injuries, future analyses should consider this category as well, especially since over 200,000 injuries are recorded annually in Italy [
4].
The social cost associated with fatalities is calculated for each black spot using Equation (4).
where
F is estimated annual fatalities (persons);
uSCF represents unitary social cost for each fatality (€/person); and
SCF is total social cost due to fatalities (€). Similar procedures can be followed to calculate the social cost for serious injuries and the total social cost for each road segment is then calculated by summing the costs for fatalities and serious injuries.
This procedure can be applied to both the current conditions of the black spot and the proposed design solutions. The economic benefit of the design alternative is determined by calculating the reduction in annual social costs through Equation (5).
where
SCsaved is annual reduction in social cost;
SCbefore represents annual social cost under current conditions; and
SCafter is annual social cost after road improvement intervention.
3.4.3. Estimation of Intervention Costs
Estimating intervention costs and assessing economic feasibility are essential to ensuring a project’s sustainability and alignment with its objectives, and it can be carried out through a parametric approach during preliminary project phases. The parametric estimation method is particularly effective for evaluating various project options quickly and efficiently. It leverages standardized parameters to estimate intervention costs without requiring detailed quantitative analysis of every component. This allows for a rapid preliminary cost assessment, aiding in selecting options that are both economically feasible and technically sound.
Parametric estimation uses indicators derived from historical data and official price lists to calculate total costs based on project quantities. Common indicators include unit costs per square meter, cubic meter, or installed unit, depending on the type of intervention. Principal cost categories considered in the parametric estimation are as follows:
Materials and labor, calculated using unit costs applied to projected quantities;
Site management costs, such as setup, maintenance, and lighting, typically calculated as a percentage of total project costs;
General project expenses covering design, project management, and other associated costs, estimated at 13–17% of total project costs [
30], with an additional 10% for contractor profit [
31].
The parametric estimation provides a reliable preliminary cost assessment, ensuring economic viability and supporting informed decision-making during early planning stages.
3.4.4. Prioritization of Interventions
By estimating benefits from reduced social costs of accidents and comparing them with intervention costs, the Benefits/Costs (B/C) ratio serves as a tool for selecting optimal project solutions.
Using the B/C ratio is rooted in the economic concept of opportunity cost, which evaluates the trade-offs of investing limited resources. This method focuses on maximizing safety benefits per unit of expenditure, directing resources toward the most impactful interventions. The B/C ratio is advantageous as it identifies solutions that provide the greatest reduction in accident-related social costs while minimizing missed opportunities.
The decision-making process performs in two stages, as follows:
Site-specific optimization, in which for each identified critical point, the intervention with the highest B/C ratio is selected as the most cost-effective solution;
Network-wide prioritization, in which across all critical points, interventions are ranked by their B/C ratios to establish a priority hierarchy, identifying sites with the greatest need for action.
Before implementing prioritized interventions, field inspections are conducted to confirm the necessity of the proposed solutions.