The Relationships Between iRAP Star Rating Score and Various Safety Performance Indicators

The iRAP Star Rating assessment has gained global recognition as an effective strategy for reducing fatalities caused by road crashes. As part of its road safety initiatives, Indonesia has also adopted the iRAP Star Rating methodology with the goal of achieving a 3-Star Rating or higher for over 75% of its road network, with a focus on prioritizing toll roads. However, it remains crucial to ascertain the effectiveness of the iRAP Star Rating methodology with regards to safety performance indicators such as crash and fatality rates and the number of crashes and fatalities. Unfortunately, there have been limited studies examining the relationship between safety performance indicators and iRAP Star Rating, with only a few conducted in Indonesia. Most of these studies only focused on iRAP Star Rating results and used a limited dataset without examining correlations to safety performance indicators. Thus, this study aims to estimate the relationship between the iRAP Star Rating Score (SRS) and safety performance indicators. The results indicate a positive correlation between SRS and safety performance indicators, with a higher SRS indicating a higher level of hazard and correspondingly higher safety performance indicators values. However, it is important to note that the R² value was not particularly high (ranging from 0.2-0.7), suggesting that this relationship may not be accurately reflected, possibly due to factors unique to the Indonesian environment and human factors not specifically accounted for in the iRAP methodology.

Crash rate; Fatality rate; iRAP star rating; Star Rating Score (SRS); Toll road

Li et al. (2024) reviewed the estimation of fatalities and serious injuries (FSI) saved by iRAP protocols in 74 countries. The study suggested more extensive applications of iRAP protocols to improve road safety as it could significantly reduce FSI through its countermeasures. Through before and after studies using crash data and iRAP protocols and its countermeasures, there were evidence on reduction of crashes in UK (iRAP, 2011), and reduction of FSI by 50% in Sweden (EuroRap, 2020; Vadeby, 2016; Carlsson, 2009).

EuroRAP (2011) conducted a study on the relationship between average crash rates and the iRAP Star Rating/Road Protection Score (RPS) in multiple European countries using various models. The findings from these studies showed variations, but overall, they suggested that higher Star Ratings were associated with a decrease in average crash rates or crash cost and vice versa. Jurewicz and Excel (2016) utilized AusRAP protocols (Austroad, 2024; Metcalfe and Smith, 2005) to calculate the relative risk of different types of severe crashes and used these results to predict the frequency of severe crashes in the Australian National Risk Assessment Model (ANRAM) (ARRB, 2015). A validation study conducted by Ambros, Borsos, and Sipos (2017) on rural roads in Hungary using a state-of-the-art empirical Bayes approach (Hauer et al., 2002; Hauer, 2001) confirmed the relationship between increasing Star Rating and decreasing crash frequencies. However, the study also found that the Star Rating had a minor influence and an unexpected positive relationship with crash frequency.

There were only limited studies regarding the relationship between iRAP Star Rating and crash rate/frequency in Indonesia. Hadi (2015) conducted a study on a 28-km length of a national road segment in Central Java, Indonesia, and found that segments with higher Star Ratings were associated with lower crash rates. However, the study only considered Star Ratings between 1 to 3, as no segments with Star Ratings 4 or 5 were included in the study. Another study by Rahmita and Malkamah (2020) examined the relationship between crash rate, number of crashes, Star Rating Score (SRS), and Star Rating band for a 10.6 km segment of National Road in Java Island. The study showed a weak correlation (R² less than 0.07) between the SRS and Star Rating band with crash rate and number of crashes. Nevertheless, none of the studies explored the relationship between iRAP Star Rating and crash severity, despite the fact that iRAP aims to evaluate road infrastructure features that impact crash likelihood and severity.

In compliance with United Nations Sustainable Development Goals (UN SDGs), one of the road safety targets stated in the Indonesian National Road Safety Plan 2021-2040 is to achieve a minimum 3-Star Rating, using iRAP, for more than 75% of Indonesia’s road network (Central Government of Indonesia, 2022). The primary focus is on enhancing road safety on toll roads, given their higher crash rates compared to other road types in the country. To achieve this goal, it is crucial to understand the relationship between crash data (such as the number of crashes, severity, crash rates, and fatality rates) and the iRAP Star Rating Score (SRS). This understanding will help determine the effectiveness of the iRAP assessment in Indonesia, which has distinct driving behavior compared to other countries. The purpose of this study is to assess toll road safety using iRAP, and estimate the relationship between iRAP Star Rating, crashes, and their severity levels in Indonesia toll road case study.

The iRAP methodology assesses the road safety in a proactive manner and identify road infrastructure deficiencies towards providing some countermeasures to improve road safety by eliminating highrisk roads (Uddin and Raihan, 2023; iRAP, 2021; Hoque et al., 2010).

iRAP Star Rating is generated using ViDA online platform for every 100-meter road segment (iRAP, 2020). Road attributes required in ViDA to calculate the SRS include roadside severity (distance and type of side object), vehicle flow, speed limit, operation speed, sight distance, intersection, pave shoulder, access point, number of lanes and lane width, grade, road condition, delineation, and the existence of shoulder rumble strip, curvature, traffic calming, service road, street lighting (iRAP, 2022). Only vehicle Star Rating was calculated in this research, as pedestrians, motorcyclists, and cyclists are not permitted to use the toll road. In addition to road attributes, the number of fatalities for each crash type is also an input in Vida for fatality estimation.

Figure 1 Study Locations

The roadside hazards in ViDA were coded according to the type of roadside object which is part of “Severity” in formulation (1), along with its distance from the edge line to the object per 100m road segment. If there were variations in road attributes within a coding segment for the length of 100m, the attribute with the highest risk would be coded for the segment. In the case of choosing the roadside object with the highest risk, iRAP has compiled a list of 62 roadside hazards, ranked according to their relative risk, to assist in identifying the objects with the highest risk. This list can be found in iRAP (2022). The top-ranked hazard on the list is the cliff, regardless of its distance from the edge line. In contrast, wire rope barriers located more than 10m from the edge line are considered to have the lowest risk. However, it is important to note that these rankings were based on research from around the world and may not necessarily reflect the specific environment of Indonesia.

Formulation (1) highlights the importance of operating speed as a factor influencing the crash type scores. This formulation considers operating speed as a combined risk factor without separating the effects of speed on the likelihood of a crash occurring and the severity of a crash. Higher operating speeds lead to higher crash-type scores, indicating a higher relative risk of serious injury and fatality for road users. The likelihood component of the formulation (1) takes into account several road attributes, including lane width, number of lanes, curvature and its quality, delineation, shoulder rumble strip, road condition, grade, skid resistance, intersection type, and its quality, the presence of street lighting, sight distance, and traffic calming. On the other hand, the severity component considers the roadside object and its distance, paved shoulder width, median type, intersection type, and access points to properties. The relationship between each attribute is based on previous studies collected by iRAP in each iRAP methodology factsheet (iRAP, 2014). Figure 2 illustrates the formulation (1) and its attributes in more detail.

SRS is apportioned to Star Rating bands to determine the Star Rating for every 100 meters of the road as in Table 1 based on iRAP. Star Rating Bands are not equally distributed. Thus, the lower the Star Rating, the greater the variance between SRS. A higher SRS indicates a more hazardous road. On the opposite, the lower the Star Rating band, the more dangerous the road. Star rating band 1 is the most dangerous, while 5 is the safest. A star rating band is commonly used to describe the star rating on the map or table because it is simpler and easier to understand (iRAP, 2024). On the other hand, SRS is the calculation for the relative risk of death and serious injury without any simplification. The distribution of SRS into Star Rating bands for vehicle occupants is as shown in Table 1:

Figure 2 SRS equation in more detail (formulation 1) (iRAP, 2014)

3.1. Star Rating Bands on Indonesia Toll Road

The star rating band was calculated using Vida based on the formulation (1), which consider the risk factor based on inputting variable for iRAP (road attributes and fatality estimation). It describes the level of road safety from 1 (highest risk) to 5 (safest road).  The Star Rating bands for Indonesia Toll Roads in the case study can be seen in Table 2 and Figure 3.

The results indicate that 67.25% of the toll road network achieved a 3-star or better rating. Among the network, only 0.19% achieved a 5-star rating, indicating the lowest risk, while 2.54% of the network received a 1-star rating, representing the highest risk. The majority of the network achieved a 2-star or 3-star rating, with 30.21% and 47.56% of the network falling into these categories, respectively. In order to improve road safety, it is important to focus on the specific areas with lower star ratings, particularly Star Ratings 1 and 2. A detailed observation of these areas can help identify the best countermeasures to address the safety issues. The Safer Roads Investment Plan (SRIP) in iRAP recommends a number of road safety measures for these segments, including adding a central median barrier, and roadside barrier, improving delineation, and adding shoulder rumble strips in several segments.

Figure 4 Wirerope barrier median (a) and wide median (b) in Cipali toll road

3.2. Number of Crashes, Crash Rates, Fatalities, Fatality Rates, and Star Rating Score

Analysis of the number of crashes per 100m segment shows that most segments had no crashes, with some having one crash and only a few with two or three crashes due to the consideration of only one year of crash data. The data were then combined into a 1km segment to allow for more variation in the number of crashes. The SRS for the 1km segment was the average SRS for the ten-100m segment.

The number of crashes was in the range between zero to four, except for one segment, which had ten crashes. The segment was KM 91-92B of the Cipularang toll road, which is one of the blackspot locations in the Indonesian toll road network. The corresponding crash rate for this segment was 79.8 crashes per 100 million vehicle-km. Figure 5 presents the distribution of the number of crashes and the crash rate for each of the SRS.

Figure 6 Distribution of the number of fatalities vs SRS (a) and fatality rate vs SRS (b)

3.3. Relationship Between SRS and Number of Crashes, Number of Fatalities, Crash Rate, and Fatality Rate

To establish the relationship between SRS and the number of crashes, fatalities, crash rates, and fatality rates, the data in Figure 5 and Figure 6 were grouped according to the number of crashes. For each number of crashes, the number of fatalities for all segments with a certain number of crashes was summed up, and the SRSs were averaged. Therefore, for example, road segments with only one crash had a cumulative fatality of 39 and an average SRS of 10.912.

Grouping the data based on the number of crashes allowed for a more accurate analysis of the data distribution. Nonetheless, the data point representing ten crashes on the Cipularang Toll Road was excluded from the analysis due to its outlier status. The resulting groups of data are thus presented in Table 4.

Table 4 Crash Data in Groups