A Comparative Study of Bikeability Index and CycleRAP in Examining Urban Cycling Facilities

Abstract

This study conducts a comparative analysis of the Bikeability Index (BI) and CycleRAP in assessing urban cycling infrastructure. The BI, developed in previous research, evaluates cycling conditions through a user-centric framework incorporating safety, comfort, attractiveness, directness, and coherence. In contrast, CycleRAP employs a data-driven methodology focusing on safety by assessing crash risks and severity across different cycling facilities. Using field data and online tools, this research applies both methods to bicycle infrastructure in Hasselt, Belgium, comparing their results and evaluating their alignment in identifying safety concerns and infrastructure needs. A significant correlation between BI and CycleRAP scores was observed, indicating that a higher bikeability score corresponds to reduced safety risks measured by CycleRAP. The study highlights the complementary nature of the two tools, emphasizing the broader insights of the BI and the focused safety evaluations of CycleRAP. The BI safety score extracted from the BI equation showed an even stronger correlation with CycleRAP, suggesting that despite using different methodologies, both indices can yield similar results. These findings provide meaningful guidance for urban planners seeking to enhance cycling infrastructure safety.

1. Introduction

With the growing emphasis on sustainable transportation, cycling has emerged as a key mode of transport in urban areas, offering significant benefits in terms of health, reduced emissions, and improved traffic management [1,2,3]. This shift towards cycling has prompted cities worldwide to invest in dedicated cycling infrastructure, including bike lanes, parking facilities, and bike-sharing programs [4,5]. As a result of these investments, there has been a noticeable increase in people choosing bicycles for their daily commutes and recreational activities [6,7]. However, the transition to a more bike-friendly urban environment is not without challenges, as it requires careful planning, public education, and sometimes contentious reallocation of road space. The increasing popularity of cycling as a sustainable mode of transportation has led to significant changes in urban landscapes. As cities adapt to this growing trend, they face opportunities and challenges in creating a more bicycle-friendly environment. While the benefits of cycling are well-documented, including improved public health, reduced carbon emissions, and alleviated traffic congestion [1,2,3], implementing cycling infrastructure requires careful consideration so that it is safe and comfortable for the riders.

Over the years, various methods have been developed to assess the aspects of safety and comfort for the cyclist on the given bicycle path or tracks. Some of these methods often focus on single aspects, such as either comfort or safety [8,9,10]. Some approaches take a more comprehensive view, evaluating the overall cycling environment by considering multiple factors, including comfort, safety, attractiveness, directness, and coherence [11,12]. Some of the well-known methods are bicycle level of service (BLOS) [13,14], bicycle safety index [9], bicycle safety index rating (BSIR) [15,16], level of traffic stress (LTS) [17,18], BI [11,19], and dynamic comfort index (DCI) [10]. Each of these methods evaluates different dimensions of the cycling experience. For instance, BLOS focuses on the performance of the cycling facilities, while LTS assesses the level of traffic stress experienced by cyclists due to vehicular interactions. These diverse methodologies offer a comprehensive toolkit for researchers and urban planners, enabling them to identify specific areas of improvement in bicycle infrastructure and to suggest solutions that enhance both safety and comfort for cyclists.

However, there are a few limitations associated with these methodologies. For example, the BSIR method developed by Sorton and Walsh [20] did not consider some important indicators, such as pavement and bicycle lane conditions [9]. Similarly, traditional BLOS methods are often criticized for not considering transport modes, like electric bikes and scooters, that utilize the same paths as bicycles [21]. Another important aspect often overlooked is generalizability since the developed methods are implemented at a much lower scale and may not apply to different situations [22]. The LTS models have also received similar criticism and are argued to be tailored to where these studies were conducted [23]. In addition, most developed methods consider the assessment at an intersection or segment level while rarely capturing variations at specific points along a route. Moreover, segment-based assessments typically aggregate conditions over a stretch of 100 to 500 m, which may obscure localized variations since they do not evaluate multiple points independently along the segment. A comfort index, such as DCI, is computed at the point level; however, such methods only consider bicycle pavement condition, which does not give an overall assessment.

The International Road Assessment Programme (IRAP) developed CycleRAP to fill this gap. The tool mainly focuses on the safety of bicycle facilities such as bike lanes, off-road bike paths, or multi-use bicycle paths. It provides data-driven insights into the quality and safety of cycling infrastructure, allowing cities to optimize their planning strategies [24]. City governments worldwide are investing in bicycle infrastructure to attract more people to biking [25,26]. However, safety remains the priority of the citizens; if it is perceived as unsafe, they are less likely to cycle [27]. CycleRAP helps urban planners and policymakers improve cycling infrastructure conditions and promote active transportation by considering factors such as road safety, connectivity, and surface quality. The CycleRAP tool has been implemented in four countries and three continents: Barcelona and Madrid in Spain, Bogotá in Colombia, Fayetteville in the United States, and São Paulo in Brazil [24]. Figure 1 shows the CycleRAP Demonstrator Tool. It provides a structured way to assess risks, helping authorities implement more effective safety measures. The CycleRAP Demonstrator Tool evaluates bicycling infrastructure safety by analyzing various facilities and assessing their risk levels based on crash likelihood and severity. It considers multiple indicators, such as facility design, traffic conditions, and vehicle interactions, to assign risk scores ranging from low to extreme, helping identify high-risk areas for cyclists.

Among assessment methods, the BI has also gained significant attention in recent years due to its holistic approach to assessing how conducive urban environments are for cycling. The BI incorporates factors like comfort and safety and considers factors such as connectivity, accessibility, and proximity to key destinations [11,28]. The concept of bikeability, which refers to how conducive an environment is to cycling, is crucial in encouraging this mode of transport [29]. The BI can be analyzed at different levels, from specific street segments and intersections to larger zones or even entire cities [28]. Several studies focused on street segments or bicycle lanes, often aggregating data from 100 to 500 m [19,30]. This assessment allows for a localized analysis of bikeability, identifying specific issues within particular stretches. Assessing smaller segments or lanes provides a detailed understanding of localized issues, allowing precise identification and targeted improvements.

In some cases, BIs were also developed for intersections and zones within cities [31,32,33]. This helps assess connectivity and safety at critical junctions where cyclists interact with motor traffic [33]. Zone-level analysis aggregates conditions within an entire neighborhood, providing a broader understanding of the area’s bikeability. A few studies conducted city-level BIs, providing a comprehensive overview of cycling conditions across the city [34,35]. This approach is better suited for macro-scale policymaking and urban planning.

This study compares the results from the BI and CycleRAP at the street level, focusing on their effectiveness in assessing cycling infrastructure. While the BI provides a holistic view of urban bikeability by considering factors like infrastructure quality, safety, and connectivity, CycleRAP uses a data-driven approach to evaluate road safety. The BI approaches developed in the past have been utilized independently to evaluate bikeability and classify the bicycling environment based on the indicators. While previous research has explored the relationship between bike-friendly infrastructure and road safety, limited studies have directly validated the developed BI against comparable tools. Existing studies typically apply the BI without assessing its reliability compared to established tools, leaving a gap in understanding its accuracy and applicability.

Additionally, this study provides empirical evidence on the strengths and limitations of the BI and CycleRAP, helping urban planners and policymakers make more informed decisions when selecting appropriate assessment tools for cycling infrastructure improvements. Hence, this study compares the results from the BI method developed in our previous research [36] and CycleRAP across different bicycle facilities. In addition, the purpose of comparing CycleRAP and the BI is to understand the relationship between the results of both tools and assess how they align in evaluating bicycle facilities, especially the safety of routes. The free online tool was utilized for this research since we had no access to other versions. The CycleRAP tool was chosen because it has been tested across five cities with varying cycling infrastructure facilities. This study includes two comparisons. First, we make a broad comparison between the BI and CycleRAP, as the BI considers multiple factors versus a safety-focused approach of CycleRAP. Some indicators overlap but are categorized differently in both methods, such as pavement condition/type being classified as comfort indicators in the BI. In contrast, it is also considered a safety indicator as a ’loose or slippery surface’ in CycleRAP. Second, we make a direct comparison between CycleRAP and the safety aspect of the BI which is extracted from the BI equation. Also, by comparing both tools, the study aims to determine how each method identifies localized issues, evaluates safety conditions, and supports targeted improvements at the street level. This comparison highlights the strengths and limitations of each tool in optimizing urban cycling environments and informing planning strategies.

2. Materials and Methods

2.1. Study Area

The study was conducted in Hasselt (capital of Limburg province), Belgium, known for its excellent bicycle infrastructure. Hasselt has a diverse bicycle infrastructure featuring various bike lanes, bicycle streets, pavement materials, and prioritization measures on the street and at the junction. Figure 2 shows examples of bicycle facilities in Hasselt. Figure 2a shows a wide, dedicated bicycle path separated from the road. It is paved and runs alongside a grassy area, providing a safe and pleasant cycling route. Figure 2b shows a narrow bike lane marked on the side of the road. The bicycle lane is separated from the motorized traffic. This is achieved by placing curbs and bicycle paths slightly higher than the adjacent road lanes. Figure 2c shows a bicycle lane on the road. The bicycle lane is clearly marked. Figure 2d shows an urban street with a bicycle lane painted yellow on both sides of the road approaching a junction. Figure 2e shows a blue circular sign indicating a bicycle lane or path. The sign has arrows pointing up and down, suggesting it is a two-way bicycle route. This signage helps direct cyclists and alert other road users to the presence of bicycle traffic. Figure 2f shows a red-colored lane (bicycle prioritized junction). The red coloring helps to distinguish and increase visibility for cyclists’ safety at the junctions. Figure 3 shows the map of the case study area, which also shows the street section IDs used for the analysis.

2.2. Calculating Bikeability Score

The BI developed in our previous study [36] is an analytical framework established based on previous research in cycling infrastructure assessment [11,33,34,37]. Our methodology employs a point-based system that evaluates cycling infrastructure through five fundamental bicycle infrastructure design principles: safety, comfort, attractiveness, directness, and coherence. The developed BI integrates quantitative measurements of infrastructure characteristics and cyclists’ perceptions, allowing for a comprehensive assessment of bicycle-friendly environments.

$$B{I}_W=j_c({\sum }_{i=1}^n{C}_{ci}{S}_{ci})+j_s({\sum }_{j=1}^m{C}_{sj}{S}_{sj})+j_a({\sum }_{k=1}^p{C}_{ak}{S}_{ak})+j_{dc}({\sum }_{l=1}^q{C}_{dcl}{S}_{dcl})$$

(1)

where:

• BI_W = bikeability weighted index;
• j_c = coefficient/weight of comfort criteria;
• j_s = coefficient/weight of safety criteria;
• j_a = coefficient/weight of attractiveness criteria;
• j_dc = coefficient/weight of directness and coherence criteria;
• C_ci = co-efficient/weight of comfort indicators;
• S_ci = score of comfort indicators;
• C_si = co-efficient/weight of safety indicators;
• S_si = score of safety indicators;
• C_ai = co-efficient/weight of attractiveness indicators;
• S_ai = score of attractiveness indicators;
• C_dci = co-efficient/weight of directness and coherence indicators;
• S_dci = score of directness and coherence indicators;
• n,m,p,q = total number of indicators in each category (comfort, safety, attractiveness, directness and coherence).

Each bicycle infrastructure design principle was weighted (e.g., comfort = 0.86, safety = 1, etc.) based on the cyclist’s preferences through Likert scale surveys in the previous study [36].

Table 1. Example of the BI calculation for Section 5 (Source: [36]).

Criteria	Criteria Weight (1)	Indicators	Indicator Notations (2)	Indicator Weights (3)	Score of Indicators (4)	Indicator Weighted Scores (5) = (3) × (4)	BIMS (6) = (3) × 1	BIW (7) = (1) × ∑(5)	BIMP (8) = (1) × ∑(6)	BI% = ∑(7)/∑(8) × 100
Comfort	0.86	Presence of bicycle infrastructure	CMF01	0.595	1.00	0.595	0.595	1.849	2.403	86.26
		Pavement type	CMF02	0.646	1.00	0.646	0.646
		Bike lane width	CMF03	0.653	0.93	0.607	0.653
		Presence of sidewalk	CMF04	0.598	0.00	0.000	0.598
		Grade	CMF05	0.302	1.00	0.302	0.302
Safety	1	Presence of bicycle infrastructure	SFT01	0.753	1.00	0.753	0.753	2.848	2.848
		Motorized traffic speed	SFT02	0.640	1.00	0.640	0.640
		Traffic control devices	SFT03	0.561	1.00	0.561	0.561
		Street lighting	SFT04	0.471	1.00	0.471	0.471
		Car parking along the cycle path	SFT05	0.423	1.00	0.423	0.423
Attractiveness	0.7	Trees/green area and landscaping	ATR01	0.477	1.00	0.477	0.477	0.674	0.674
		Bicycle parking	ATR02	0.486	1.00	0.486	0.486
Directness and Coherence	0.76	Presence of cycle facilities at a traffic signal	DC01	0.658	1.00	0.658	0.658	0.754	1.176
		Road signage	DC02	0.555	0.00	0.000	0.555
		Interruptions (i.e., pedestrian crosswalks, intersections, or bus stops on bicycle lanes)	DC03	0.334	1.00	0.334	0.334

shows an example of the BI calculation, where various criteria, such as comfort, safety, attractiveness, and directness and coherence, are evaluated. There are multiple indicators under each criterion, each one having different weights and scores. The score for each indicator is assigned to the bicycle facility based on their presence, partial presence, or absence. In the study [36], the survey response was also used to estimate the score of each indicator using Min–Max normalization to ensure consistency across different indicators. The scores of the indicators are provided in Appendix A.

Each indicator also has a specific weight based on the cyclist’s perception. For instance, under “Comfort”, five indicators (CMF01 to CMF05) are weighted according to their coefficients (e.g., CMF01 = 0.595, CMF02 = 0.646). The weighted coefficient of the indicators is calculated by multiplying the coefficient of the indicator with the observed scores. For example, the weighted score of CMF01 can be calculated as 0.595 (coefficient) x 1.00 (indicator score based on the bicycle lane’s presence) = 0.595. The BI_W is calculated as the sum of the indicators’ weighted scores multiplied by the respective criterion weight. For instance, the comfort criteria weightage score is 1.849, calculated by adding the weighted scores under the criteria. The BI_MP is completed in the next step by adding BI_MS for each criterion. Finally, the percentage score (BI%) is calculated by dividing its weighted score (BI_W) by the total maximum possible weighted score for that criterion (BI_MS), then multiplying by 100. The resultant bikeability index score can be categorized into five levels: A (81–100) Extremely Bikeable, B (61–80) Bikeable, C (41–60) Fairly Bikeable, D (21–40) Less Bikeable, and E (0–20) Not Bikeable, with each category indicating the extent of improvements needed for better cycling infrastructure [36].

2.3. Calculating CycleRAP Score

This study utilizes the online CycleRAP tool (https://irap.org/cyclerap/demonstrator/, accessed on 7 November 2024) to compute the CycleRAP score for the streets in the study area. The CycleRAP focuses on assessing bicycling infrastructure safety by evaluating the risk and severity of potential crashes for cyclists and light vehicle (LV) users, like e-scooters and cargo bikes [24]. In addition, it aims to promote safer urban mobility by identifying risk factors that contribute to bicycle crashes and determining how severe those crashes might be. We have used the detailed methodology provided by CycleRAP on the website. The methodology provides a guide on how each indicator should be recorded. For example, facility access would be inadequate if, due to congestion, obstructions, or other issues, some or all bicyclists and light vehicle users are forced to use alternative routes (e.g., adjacent paths or roadway). Similarly, the flow direction attribute shows the flow of the bicycle along the bicycle facility to see whether it is in one direction or two.

This CycleRAP method uses a three-part structure, considering crash initiators (e.g., avoiding obstacles, speed, or vehicle impacts), contributing factors (e.g., road conditions, limited space), and severity determinants (e.g., vehicle speed, road hazards) [38]. In the CycleRAP method, crash initiators are triggered when certain conditions or attributes are detected, such as the presence of obstacles, tram rails, vehicle interactions, etc. These triggers are combined with factors that increase the likelihood of a crash, and the severity of the crash is then calculated based on specific conditions present on the bicycle routes, for example, the vehicle’s speed or heavy vehicle flow, etc., on the lane next to the bicycle facility.

The overall risk score is determined by adding the severity of different crash types, such as vehicle–bicycle (VB), bicycle–bicycle (BB), bicycle–pedestrian (BP), or single-bicycle (SB) crashes. Risk scores are categorized into low, medium, high, and extreme levels. Incorporating various mobility options, CycleRAP also evaluates newer micro-mobility trends, recognizing how different vehicle types and speeds impact crash likelihood and severity. It includes evidence-based risk factors from global studies on cyclist safety, such as road design, facility width, and interaction points like intersections.

Table 2. CycleRAP scores and bands.

Crash Type	Low Risk (Green)	Medium Risk (Yellow)	High Risk (Red)	Extreme Risk (Dark Red)
VB	<10	10 to 25	25 to 60	>60
BB, BP, SB	<5	5 to 10	10 to 20	>20

shows the risk associated with values for all four crash types. The overall CycleRAP score is also labeled as a low, medium, high, or extreme risk if at least one of the individual crash types falls within that category. The VB crash values are higher than the other three crash types (BB, BP, and SB) mainly due to the severity and likelihood of these crashes. VB crashes involve motor vehicles, which are much heavier and faster than bicycles. This makes these crashes far more dangerous and likely to cause serious injuries or fatalities compared to BB, BP, or SB crashes. Higher speed and traffic volume increase risk. The CycleRAP model includes risk multipliers for vehicle speed and traffic volume. As more cars and trucks are present, the risk of VB crashes increases significantly. The CycleRAP score aggregates them all to compute the score. In addition, the values are transferred to the band (colors), which then tells the risk in the given section.

2.4. Calculating the BI Safety Score

The BI safety score was calculated similarly to the BI score but only with the safety indicators in

Table 3. Indicators for the BI and the data collection method.

Criteria	Indicator Notations	Measurement Criteria	Data Collection
Comfort	CMF01	Solitary bike path.	Google Maps
		Physically separated (by height or space) bicycle lane.
		Bicycle lane.
		Bicycle-prioritized streets.
		Suggested bicycle path.
		Bicycle path shared with motorized traffic.
	CMF02	Asphalt-paved.	Google Maps
		Concrete-paved.
		Paving slabs.
		Cobblestones.
	CMF03	Unidirectional, wide (≥2 m).	Field visit
		Unidirectional, narrow (<2 m).
		Double direction, wide (≥3 m).
		Double direction, narrow (<3 m).
		Shared.
	CMF04	Buffered from cycle lane.	Google Maps
		Adjacent to cycle lane.
		Shared with cyclist.
	CMF05	Low (1–3%).	GIS
		Medium (3–6%).
		High (>6%).
Safety	SFT01	Solitary bike path.	Google Maps
		Physically separated (by height or space) bicycle lane.
		Bicycle lane.
		Bicycle-prioritized street.
		Suggested bicycle path.
		Bicycle path shared with motorized traffic.
	SFT02	30 km/h.	Field visit
		50 km/h.
		70 km/h.
	SFT03	Availability of traffic signals at intersections.	Field visit
		Non-availability of traffic signals.
	SFT04	Good street lighting (not exceeding 60 m apart from one another).	Field visit
		Limited street lighting (the distances between the light poles are longer than 60 m).
		No street lighting.
	SFT05	No car parking.	Google Maps
		Car parking with a buffer area.
		Car parking without a buffer area.
Attractiveness	ATR01	Bicycle route/lane along trees and landscaping or water area.	Google Maps
		Bicycle route/lane without trees and landscaping or water area.
	ATR02	Parking facilities at key destinations (e.g., shops, stations, etc.).	Field visit
		No parking facilities at key destinations (e.g., shops, stations, etc.).
Directness and Coherence	DC01	Presence of bicycle facilities at intersections.	Google Maps
		Partial presence of bicycle facilities at intersections.
		Non-presence of bicycle facilities at intersections.
	DC02	Well-signposted.	Field visit
		Partially signposted/signage missing at key location.
		No signage available.
	DC03	One or no interruptions.	Google Maps
		Two or more interruptions.

. This helps assess how the safety score correlates with CycleRAP. The safety indicators in the BI include indicators directly associated with safety metrics, such as motorized traffic speed, car parking along bicycle paths, and traffic signals at the junctions. In Equation (1), the safety score is extracted as a primary factor, ensuring that only data relevant to predefined safety criteria contribute to the score. The safety score can then be calculated using Equation (2), derived from the measurement criteria identified in Table 3 and the weight of indicators in Table 1.

$${\mathrm{BI}}_{\mathrm{s}}=j_s({\sum }_{n=1}^x{C}_{si}{S}_{si})$$

(2)

2.5. Data Collection

The data for this study were collected at the street level within the case study area, utilizing two primary methods: Google Maps and field observations. These methods were employed to gather data relevant to the BI and CycleRAP assessments. Google Maps (Street View, SV) was used to extract geospatial information and assess street characteristics such as the presence of bike lanes, sidewalks, pavement type, etc. For data collection, the date of the SV image was checked to ensure it was recent. If the SV image was older than three years, a field visit was still carried out to confirm the current condition of the street. If measurements or counting were involved, such as measuring widths or traffic flow counting, field visits were conducted. Similarly, field visits were performed if certain features, such as street lighting, could not be identified from Google Maps SV.

Observations included assessing the quality of bike lanes, surface conditions, road safety features (e.g., lighting, signage, and lane separation), factors affecting comfort and safety, and potential obstacles. Table 3 shows the indicators, their notation, and possible options for each indicator (measurement criteria) on bicycle facilities or streets. Each indicator is measured based on the options listed under measurement criteria, which are the possible options that a bicycle street or facility can have.

Table 4. Indicators for CycleRAP and the data collection methods.

Facility Type	Indicator Measurement	Data Collection
Facility access	Adequate.	Google Maps
	Inadequate.
Loose or slippery surface	Not present.	Field visit
	Present.
Tram or train rails	Not present.	Google Maps
	Present.
Major surface deformation or drain	Not present.	Field visit
	Present.
Fixed obstacle on facility	Not present.	Google Maps
	Present.
Non-fixed obstacle on facility	Not present.	Field visit
	Present.
Delineation	Not present	Field visit
	Present.
Light segregation	Not present.	Google Maps
	Present.
Facility width per direction	Very narrow (less than 1 m).	Field visit
	Narrow (1 to 2 m).
	Wide (more than 2 m).
Flow direction	One-way.	Field visit
	Two-way.
Width restriction	Not present.	Field visit
	Present.
Adjacent road lane 0–1 m	Not present.	Google Maps
	Present.
Adjacent vehicle parking 0–1 m	Not present.	Google Maps
	Present.
Adjacent severe hazard 0–1 m	Not present.	Field visit
	Present.
Adjacent object or level change 0–1 m	Not present.	Field visit
	Present.
Adjacent sidewalk 0–1 m	Not present.	Google Maps
	Present.
Adjacent road lane 1–3 m	Not present.	Google Maps
	Present.
Adjacent vehicle parking 1–3 m	Not present.	Google Maps
	Present.
Adjacent severe hazard 1–3 m	Not present.	Field visit
	Present.
Adjacent object or level change 1–3 m	Not present.	Field visit
	Present.
Adjacent sidewalk 1–3 m	Present.	Google Maps
	Not present.
Grade	<5 degrees.	GIS
	≥5 Degrees.
Curvature	No sharp turn present.	Google Maps
	Sharp turn present.
Street lighting	Present.	Field visit
	Not present.
Pedestrian crossing	Present.	Google Maps
	Not present.
Intersecting bicycle facility	Present.	Google Maps
	Not present.
Intersection approach	Present.	Google Maps
	Separate/NA.
Intersection or road crossing	Present.	Google Maps
	Not present.
Crossing facility	Present.	Field visit
	Present/NA.
Number of lanes—adjacent road	One per direction/NA.	Google Maps
	>one per direction.
Number of lanes—intersecting road	One per direction/NA.	Google Maps
	>One per direction.
Property access	Present.	Google Maps
	Not present.
Peak pedestrian flow along or across	Low.	Field visit
	Moderate to high.
Peak bicycle/LV traffic flow	Low.	Field visit
	Moderate to high.
Obs. proportion of cargo bikes	Low.	Field visit
	Moderate to high.
Bicycle/LV speed—average	<20 km/h.	Field visit
	≥20 km/h.
Bicycle/LV speed differential	<10 km/h.	Field visit
	≥10 km/h.
Road AADT	0–40,000.	Secondary data
Heavy vehicle flow	Low.	Field visit
	Moderate to high.
Road operating speed (mean)	10–140 km/h.	Field visit

shows the indicators and measurement criteria for the CycleRAP tool. Each indicator is measured based on whether it is present or absent on or along the bicycle facility. Few indicators have more than two options, such as heavy vehicle flow, facility width per direction, or road operating speed (mean). To ensure an accurate assessment of infrastructure conditions, multiple points were evaluated along each bicycle path whenever there was a change in bicycle facility characteristics. This approach captures variations in design, surface quality, and safety features, preventing reliance on a single-point assessment.

3. Results

3.1. Bikeability and BI Safety Scores

Based on Table 1, the BI scores are calculated for the case study streets.

Table 5. Comparing the BI, BI safety score, and CycleRAP scores across the streets.

Street Name	Street/Bicycle Facility Section ID	BI Score	BI Safety Score	CycleRAP Score
Maastrichtersteenweg	Section 1	69.27	2.486	20.36
Sint-Truidersteenweg	Section 2	55.82	2.067	24.36
Koning Boudewijnlaan	Section 3	83.00	2.697	13.88
Kempische Steenweg	Section 4	60.99	2.274	18.09
Inner Ring Hasselt	Section 5	86.26	2.848	12.76
Inner Ring (Outer Side)	Section 6	69.74	2.697	16.34
Witte Nonnenstraat	Section 7	55.03	2.197	20.24
Badderijstraat	Section 8	49.65	1.803	24.25
Minderbroedersstraat	Section 9	66.38	2.066	19.84
Dorpstraat	Section 10	57.35	2.489	17.93
Dokter Willemsstraat	Section 11	69.96	2.489	19.78
Lombaardstraat	Section 12	65.13	2.066	20.70
Walputstraat	Section 13	69.67	2.848	11.50
Onze-Lieve-Vrouwstraat	Section 14	73.39	2.399	17.28
Ridder Portmansstraat	Section 15	68.71	2.066	25.61
Havermarkt	Section 16	79.05	2.066	19.84
Melderstraat	Section 17	80.80	2.592	12.83
Luikersteenweg	Section 18	82.87	2.549	14.53
Isabellastraat	Section 19	66.55	2.037	22.90
Persoonstraat	Section 20	66.70	2.095	22.04
Guido Gezellestraat	Section 21	79.83	2.066	17.83
Capucienenstraat	Section 22	76.12	2.018	20.38
Sint-Jozefsstraat	Section 23	85.09	2.489	17.27
Koning Albertstraat	Section 24	77.47	2.848	14.42
Ridderstraat	Section 25	72.10	2.489	14.74
Cellebroedersstraat	Section 26	75.27	2.066	15.46
Kapelstraat	Section 27	72.75	2.489	17.97
Diesterstraat	Section 28	78.38	2.489	13.93
Aldestraat	Section 29	78.86	2.066	16.57
Schrijnwerkersstraat	Section 30	82.52	2.848	9.51

shows the BI score calculated for each street. The BI scores across different street sections vary in the case study. Section 5 has the highest BI score, 86.26, while Section 8 records the lowest, 49.65.

Table 5 also shows the BI safety scores for the study areas’ bicycle facilities, i.e., bicycle lanes, bicycle paths, or bicycle streets. Section 30, Section 24, and Section 5 (See Table 5 for section IDs) received high scores in the BI safety. These high safety scores indicate that these have well-developed bicycle facilities, with features that likely contribute to safer conditions for cyclists. In contrast, Section 8, Section 26, and Section 22 received lower BI safety scores.

3.2. CycleRAP Scores

Figure 4 and Figure 5 show how different bicycle facilities would score under varying conditions. Figure 4 shows a CycleRAP risk assessment for the cycling lane of Section 17. The overall CycleRAP score for the bicycle facility resulted in a score of 12.83 based on the facilities available for cyclists. The CycleRAP risk scores for different crash types were recorded, which were low for BB, BP, VB, and SB, and the scores of these crash risks are 2.16, 4.32, 4.19, and 2.16, respectively.

Figure 5 shows the CycleRAP score at a segment on Section 1, with an overall score of 31.75, indicating a moderate level of risk for cyclists in this area. The figure illustrates how the absence of more indicators impacts the CycleRAP score. The street has a separate bicycle lane from motorized traffic; however, risks for BP, VB, and SB incidents are rated as medium. Few indicators lead to an increase in the CycleRAP score and the risk of BP (8.03), VB (11.66), and SB (8.03) crashes. Few design elements and traffic conditions contribute to this CycleRAP score. For instance, the bicycle lane is used for two-way bicycle movement, but its width is narrower than the recommended 2 m per direction. This narrower width limits maneuverability and makes it challenging for cyclists to safely pass each other, thereby increasing the risk of crashes. Additionally, car parking is allowed on the street, which creates safety hazards, such as dooring incidents and restricted visibility for cyclists. Other contributing factors include intersecting cycling facilities and pedestrian crossings on the bicycling lane, which may lead to unexpected stops or unanticipated movements, further elevating safety concerns for cyclists.

3.3. Comparison of BI and CycleRAP Scores

Table 5 shows the case study streets’ BI, BI safety, and CycleRAP scores. To compare the scores, it was necessary to conduct appropriate statistical tests to evaluate their relationship. Before selecting the appropriate test, we checked the normality of the BI scores, BI safety scores, and CycleRAP scores and assessed them using the Shapiro–Wilk test. The normality was also examined through histograms, Q-Q plots, and visual inspection. The results revealed that BI safety scores deviate significantly from normality. In contrast, the CycleRAP and BI scores showed no significant deviation from normality. This suggests a normal distribution for both scores.

These findings indicate that BI safety scores require non-parametric methods, while parametric approaches are suitable for CycleRAP and BI scores. Hence, we performed a Pearson correlation test to find a correlation between BI scores and CycleRAP scores and a Spearman correlation test to find a relationship between BI safety scores and CycleRAP scores. The CycleRAP and BI scores test yielded a Pearson correlation coefficient of −0.67 with a p-value of <0.001. This shows a strong inverse relationship between the BI and CycleRAP, suggesting that as bikeability increases, safety risks tend to decrease, as assessed by CycleRAP. The p-value is very low, indicating that the correlation is statistically significant, and the relationship observed is unlikely due to random chance. We performed the Spearman correlation test between the safety score (from the BI) and CycleRAP. The Spearman correlation coefficient was −0.79, with a p-value of < 0.001. The test result reveals a stronger correlation between the BI safety score and the CycleRAP assessment. It shows that increases in the safety aspect of the BI are strongly associated with decreases in safety risks assessed based on CycleRAP. The higher value of this correlation coefficient (0.79 compared to 0.67) suggests that the safety score of the BI captures aspects of safety that are more closely aligned with the CycleRAP better than with the BI in general.

The streets with higher BI scores indicate that most facilities are available to cyclists. Similarly, the lower CycleRAP score indicates the same thing; if the score is lower, the bicycle infrastructure has major facilities available to be safe. Hence, the correlation (inverse) means streets with higher BI scores tend to have lower CycleRAP scores, both indicating safer conditions for cyclists. In contrast, streets with lower BI scores exhibit higher risk levels. A similar strong correlation is observed between the BI safety score and CycleRAP for the street sections of the case study area.

Figure 6 shows the bicycle facilities’ BI, BI safety, and CycleRAP scores in the case study area. The street segments are arranged based on ascending BI scores. Figure 7 presents various street sections from the case study: (a) Section 5, (b) Section 1, (c) Section 3, (d) Section 4, (e) Section 27, (f) Section 2, (g) Section 8, (h) Section 15, and (i) Section 13. There were notable similarities and differences in the score for bicycle streets with the same infrastructure facility, i.e., cycle lanes. In general, higher BI scores correspond to lower CycleRAP scores, indicating safer cycling conditions, while lower BI scores are associated with higher CycleRAP scores, reflecting increased risk. For example, Section 5 and Section 1 have bicycle lanes physically separated from motorized traffic; however, the CycleRAP and BI scores vary considerably. Despite having a cycle lane, Section 5 was rated safer than Section 1, as it has a higher BI score (86.26 vs. 69.27), a higher BI safety score (2.848 vs. 2.486), and a lower CycleRAP score (12.76 vs. 20.36).

Several factors contributed to these different scores. Section 5 scored higher because the bicycle lane width is wider. Secondly, it is separated from motorized traffic by more than 2 m, which is not the case with Section 1. Light segregation exists between the bicycle lane and motorized traffic in Section 1; however, an adjacent road within 1 m poses more risk to riders than in Section 5. In addition, some other features are no vehicle parking, no road intersects the bicycle lane along most of the lane, and the two-way bicycle lane is delineated. These factors are considered in the BI and CycleRAP, as seen in another bicycle lane in Section 3, which has a higher BI score (83.00) and a lower CycleRAP score (13.88). The bicycle lane is physically separated from motorized traffic with light segregation, and delineation with the footpath is also segregated at most part of cycle lane. Section 5 and Section 3 highlight that a higher BI is associated with safer cycling conditions.

Similarly, there were cases where the BI and CycleRAP scores were in the medium range yet showed agreement. For instance, Section 4 (BI score = 60.99, CycleRAP = 18.09) showed a decent BI score with a moderate safety risk level based on the CycleRAP. The BI score is lower due to the lane being narrow and lacking physical segregation, both of which negatively impact the score. These two factors also contribute to the moderate CycleRAP score. Another factor contributing to a moderate score is the unavailability of parking facilities for a bicycle at key destinations (e.g., shops and cafes). The number of lanes (two lanes, two-way) and the AADT of the adjacent road contribute to the moderate CycleRAP. Section 27 is another example of a similar correlation (moderate BI score = 72.75 and CycleRAP score = 17.97). Section 27 is a street in the inner city where bicyclists share the street with motorized vehicles; however, bicycles are prioritized over motorized vehicles. A few factors contributing to these scores are the availability of vehicle parking on the street (without a buffer in case of the BI). In the case of the BI, the bicycle-prioritized streets receive a lower score (0.63) compared to bicycle lanes of physically separated bike paths. In addition, the attractiveness factor, i.e., the presence of trees along the route, also contributes to a higher BI, which was not the case in Section 27.

Section 2 also had a low BI score (55.82) and CycleRAP score (24.36). A few reasons for having low scores are that the bicycle lane is next to motorized traffic without physical separation (light segregation). Vehicle parking is permitted next to the bicycle lane, further contributing to the lower score. In addition, the average mean speed and annual average daily traffic were also higher along the road, which is another reason for the comparatively lower CycleRAP score. Moreover, although the bicycle lane is one-way, the lane’s width is narrow.

Similarly, Section 8 is another example of a lower BI (score = 49.65) and CycleRAP (score = 24.25). The lower BI score of Section 8 is due to bicyclists sharing the street with motorized traffic. In addition, the street is paved with cobblestone, significantly reducing the cyclist’s comfort compared to other pavement types such as asphalt- or concrete-paved. The examples of Section 2 and Section 8 suggest that a lower BI score on the bicycle facility is associated with a higher CycleRAP score. This indicates that the streets are less safe for cyclists. The BI safety score also had low scores of 2.067 and 1.803 for Section 2 and Section 8, respectively.

There were a few examples where the bicycle facilities had higher BI scores and lower CycleRAP scores. One example is Section 15, which has a moderate BI score (68.71) and a higher CycleRAP score (25.61). The comparatively higher CycleRAP score is due to the absence of delineation, light segregation between the cyclist and motorized traffic, and moderate to high pedestrian flow on the street during peak hours. The BI score is also moderate but decent, as the mentioned factors are not considered in the calculations. Another example is Section 13, which had a moderate BI score (69.67), while the CycleRAP was extremely low (11.50). The moderate BI score was due to factors the BI considers which are not considered in CycleRAP, such as road signage availability and interruption along the bicycle path or street.

Comparing the BI safety and CycleRAP scores across various streets showed a strong relationship. A lower BI safety score and high CycleRAP score imply that the bicycle facility is unsafe. Section 2 (BI safety score = 2.067, CycleRAP score = 24.36) reflects a lower safety score in the BI safety score and a relatively high CycleRAP score, indicating increased risks for cyclists. The lack of physical separation from motorized traffic, narrow lane width, and parking adjacent to the lane contribute to these scores. Similarly, Section 8 (BI safety score = 1.803, CycleRAP score = 24.25) shows a low BI safety score and a high CycleRAP score. This is largely due to cyclists sharing the street with motorized vehicles. These examples suggest low BI safety scores and higher CycleRAP scores often indicate cycling environments with limited protection and increased risk.

On the other hand, streets such as Section 5 (BI safety score = 2.848, CycleRAP score = 12.76) demonstrate a higher BI safety score category alongside a lower CycleRAP score, corresponding to safer cycling conditions. The Inner Ring bicycle lane is physically separated from traffic by more than 2 m, with a wider width and minimal vehicle intersections, contributing to high BI scores and lower safety risks. Similarly, Section 13 (BI safety score = 2.848, CycleRAP score = 11.50) follows a similar trend with a high BI safety score and a very low CycleRAP score. The street’s clear delineation, separation from motorized traffic, and minimal interruptions result in safer conditions for cyclists. These cases highlight that high BI safety scores correspond to low CycleRAP scores.

4. Discussion

Bicycle infrastructure is important in making bicycle transport more attractive [28]. Bicycle infrastructure indicators are crucial to making bicycle facilities safe [9]. Researchers have suggested various methods to assess bicycle infrastructure facilities for their safety, comfort, or overall bikeability [9,11,32]. This study compares the BI with CycleRAP to examine their similarities and differences.

One study argues that safety and security are considered the most crucial aspects for cyclists and non-cyclists in bicycling facilities [11]. Safety was considered the most important aspect based on the perception of cyclists when weighing the importance of criteria in developing the BI methodology. The BI considers the presence of bicycle infrastructure, motorized traffic speed, traffic control devices, street lighting, and car parking indicators as part of the safety score. These factors are well-reported in bicycle safety research [21,39]. These indicators are incorporated into the BI and CycleRAP using various bicycle facility options. While both indices account for the presence of car parking, considering the impact of buffer zones between parking and bicycle lanes in reducing crash risk. When vehicle parking is located with a buffer, it improves the score, indicating a safer design. Findings from this study indicate that buffer zones contributed to improved scores, reinforcing previous research that highlights their role in minimizing dooring crashes and enhancing cyclist safety [40,41].

One study suggests that bicycle-friendly cities (higher bikeability) have lower crash fatality rates for all road users, likely due to slower vehicle speeds and safer street designs [41]. This study’s findings align as theresults show a high Pearson correlation between the two methods. The higher BI score corresponds to lower CycleRAP scores and vice versa. This implies that bicycle facilities (with higher bikeability) are safer. Both methods found a strong link between bike-friendly designs and the outcomes. This relationship is well-reported in the literature, as the cities with high bicycling rates are safer for all road users due to a greater prevalence of bike facilities, particularly protected and separated bike facilities [42,43,44].

There were very few instances where the BI score was low, but the CycleRAP score was comparatively higher. This difference highlights specific cases where the correlation, though significant and high overall, did not align perfectly. The main reason is that the BI considers other factors, such as comfort, attractiveness, directness, and coherence. In contrast, CycleRAP prioritizes safety, emphasizing the contribution of specific indicators to safety. Researchers and city authorities have argued that the five principles of bicycle infrastructure design should be considered when planning bicycle facilities [11]. The weightage computed based on the survey also plays a role in the overall bikeability. Researchers argue that the weights should be considered when computing such metrics [45]. The weightage is calculated based on the cyclist’s perception as cyclists were asked to give rank based on their safety, comfort, and other BI criteria. The CycleRAP calculates the score based on the likelihood of a crash occurring and the possible severity of the crash based on the indicators. In addition, some indicators that CycleRAP considers are also considered in the BI in different categories than safety. For example, the condition of bicycle pavement described based on pavement type was considered in the comfort category. The CycleRAP considers the pavement condition under loose or slippery surface indicators, measuring it based on whether it is present or absent. These indicators can also increase safety but are more important for comfort. Our rationale for considering them in another category was inspired by the literature, as most of studies consider it as impacting comfort.

The BI metric we developed uses the approach suggested by studies, including the perception and measuring indicators for assessing bicycle facilities [11]. Evaluating the quality of bicycle networks should involve multiple factors, and the varying preferences of different users should be considered concerning these factors [46]. Other research also recommends the same approach by taking the indicator’s importance in calculating bicycle street assessment metrics [47]. Cyclists view the characteristics of a bicycle-friendly physical environment as crucial elements of bikeability, and a supportive community framework enhances the overall bikeability of the community [48]. Hence, their perception is vital when assessing bikeability.

The BI method incorporates cyclists’ preferences when weighing the indicators, whereas the CycleRAP model is evidence-driven and does not consider subjective preferences. The CycleRAP model emphasizes the integration of crash likelihood and severity determinants. Also, CycleRAP assesses safety based on a comprehensive list of forty indicators that can affect the safety of cyclists. Some of these indicators are not considered in the BI model, such as pedestrian flow, number of adjacent road lanes, property access, heavy vehicle flow, etc. Another difference between both models is the subcategories of the indicators. For instance, the bike lane width is measured in three categories (very narrow, narrow, and wide) using the CycleRAP methodology, while the BI considers four subcategories: unidirectional narrow, unidirectional wide, bidirectional narrow, and bidirectional wide. Other important indicators, such as pavement type, should be considered when suggesting a model only for bicycle safety, like the CycleRAP method. In our BI methods, these were included but in a different category. The weightage of these indicators in the model would be different when considered as a safety indicator instead of comfort.

The BI safety score, extracted from the BI, was compared separately with CycleRAP to assess their alignment in measuring safety aspects. The findings demonstrate a stronger correlation compared to that between the BI and CycleRAP. This suggests that both indices capture key safety-related factors with different methodologies. BI safety integrates safety indicators within a broader bikeability assessment, whereas CycleRAP strictly evaluates crash risks. The higher correlation between BI safety and CycleRAP compared to the BI and CycleRAP indicates that BI safety better reflects safety risks identified by CycleRAP.

This study shows that both tools can be used to assess bicycle infrastructure, but which one to use depends on the city’s needs. If safety is a major concern, CycleRAP is the better tool because it focuses on crash risks and safety factors. If a city wants to encourage more people to cycle, then the BI is the better option because it looks at safety, comfort, attractiveness, directness, and coherence. These tools help cities find problem areas in cycling infrastructure. However, a big issue is data collection; if the data are unavailable, collecting them manually takes much time. This can slow down the process. Using new technologies like drones, AI, and computer vision can help speed up data collection and make these tools easier to use in the future.

One concern of the BI is that it provides a score that can be categorized as bikeable level A-E; if plotted on a map, the BI score or category alone will not distinguish whether a street lacks safety, connectivity, or comfort. However, it can show the city’s bikeability status, but it does not show which specific factors contribute to a lower score. Moreover, another general limitation of the study is that since the study is conducted in a city with advanced bicycle infrastructure, the results might not reflect the conditions and challenges faced in cities with less-developed cycling networks.

5. Conclusions

This study has provided a comparative analysis of two methods for assessing cycling infrastructure: the BI and CycleRAP. By focusing specifically on the safety score of the BI, we have ensured a more direct and fair comparison with CycleRAP, which only measures safety. The results demonstrate a statistically significant correlation between the two metrics (BI and CycleRAP), indicating that streets with higher BI scores generally exhibit lower CycleRAP scores, signifying safer conditions for cyclists. This close correlation suggests that improved bikeability, as assessed through the multiple factors, often aligns with reduced safety risks as quantified by CycleRAP.

The findings reveal that the BI safety score aligns more closely with CycleRAP than the general BI score. This further validates that safety indicators considered within the BI contribute significantly to safer bicycle infrastructure. This relationship highlights that while the BI’s comprehensive approach includes factors like comfort, directness, and attractiveness, the prioritization of safety within the BI resembles the safety evaluations of CycleRAP. These results confirm that despite different scopes and measuring procedures, both tools effectively capture safety elements critical to cycling infrastructure.

Some differences were found in some cases where CycleRAP scores indicated moderate risk despite higher BI ratings and vice versa. These variations are primarily because the BI’s broader assessment scope considers factors beyond safety. In addition, it also considers the weightage based on the perception of cyclists. This disagreement (in some cases) underscores the complementary nature of the two tools: while CycleRAP’s deeper focus is valuable for pinpointing direct safety risks, the BI’s holistic approach identifies broader environmental enhancements that could foster a more enjoyable and supportive cycling experience. In practical terms, this study underscores the need for urban planners to consider both tools when designing and evaluating cycling infrastructure. Both tools are very useful for city planners. If the purpose is to improve bicyclist safety, CycleRAP can be utilized to assess bicycle facilities and prioritize enhancements to ensure safer conditions for cyclists. Similarly, the BI can help city authorities assess and improve the overall cycling environments that promote comfort, connectivity, and attractiveness—factors vital to promoting a sustainable cycling culture.