The Impact of Speed Bumps on Traffic Flow Speed in Urban Road Networks

Abstract

Traffic safety is a fundamental element of urban mobility, and speed bumps remain one of the most widely used measures for reducing vehicle speeds on local streets. This study investigates how different types of speed bumps influence traffic flow speed in the City of Zagreb, Croatia. A total of 208 locations were surveyed across all city districts, where geometric characteristics, regulatory compliance, and local contextual features were recorded. In addition, UAV monitoring was conducted at eight representative sites, capturing 906 vehicle trajectories and enabling the extraction of speed profiles at two measurement points per location, together with vehicle classification. This combined approach—integrating UAV-based speed tracking with a detailed geometric compliance assessment—provides a novel and reproducible methodological framework for evaluating vertical traffic-calming measures under real operating conditions. The results show substantial differences in performance between bump types. Raised platforms reduced vehicle speeds by up to 53%, while narrow platforms achieved reductions of up to 49%. In contrast, modular rubber elements exhibited noticeably weaker performance, particularly for heavy vehicles. These findings differ from previous research that has primarily focused on single bump types or limited samples and reported mixed effectiveness depending on height and material. The reductions observed in this study are operationally relevant, as they indicate which bump designs reliably maintain speeds below the 40 km/h safety threshold required on residential streets and around schools. By linking bump geometry and compliance with actual driver behaviour, this study offers a practical and transferable framework that can support urban traffic-safety planning, standardization of vertical calming devices, and improved selection of appropriate measures for mixed-traffic urban environments.

1. Introduction

Traffic safety is a critical concern in urban environments, where speed-related accidents pose a significant risk to all road users.

Speed remains one of the key contributing factors to traffic crashes. In the Republic of Croatia, inappropriate or excessive speed accounted for 18.4% of all fatal crashes [1], while in the City of Zagreb, more than 30% of all recorded traffic violations in 2024 were related to speed. In the same year, 1428 crashes in Zagreb were attributed to speed as the primary contributing factor. These figures confirm that speed management is a persistent challenge in dense urban areas and highlight the need for effective and well-designed traffic-calming measures. Although speed bumps are widely used for this purpose, their effectiveness across different designs and under mixed traffic conditions, such as those in Zagreb, remains insufficiently quantified.

Speed bumps are widely applied physical measures designed to reduce vehicle speeds and enhance safety [2]. The urban road network of Zagreb includes more than 2800 speed bumps of various designs and materials, highlighting the need for a systematic assessment of their effectiveness and regulatory compliance. The primary aim of this study is to evaluate the technical compliance, geometric characteristics, and speed-reduction effectiveness of different bump types in urban settings.

Previous research has extensively documented the role of speed bumps in improving road safety. Studies indicate that their installation in residential areas can significantly reduce the frequency and severity of traffic accidents, especially in pedestrian zones. Nevertheless, speed bumps are also criticized for disrupting traffic flow and causing discomfort to drivers and passengers [3]. This dual effect underscores the importance of assessing the trade-offs involved in their design and placement. Similar results were reported by Antić et al. [4], who analyzed the impact of speed bump height (3, 5, and 7 cm) on vehicle speeds in Belgrade. Their findings showed that higher bumps (5 and 7 cm) significantly reduced speeds and improved pedestrian safety, while lower bumps (3 cm) were insufficient to achieve the desired effect.

Speed bumps, therefore, represent an important element of broader traffic-calming strategies, which commonly combine physical measures, policy instruments, and public-awareness campaigns. This study contributes to the literature by integrating a geometric assessment of bump dimensions, a regulatory compliance analysis, and UAV-based speed measurements into a unified methodological framework that can be applied in other mixed-traffic urban environments.

Although numerous studies have examined speed bumps, many have relied on small samples or focused solely on geometric design or isolated speed measurements. Recent studies from Europe, Asia, and North America indicate that the interaction between bump geometry, vehicle type, and network-level dynamics remains underexplored, particularly in heterogeneous traffic streams. Furthermore, few studies integrate compliance evaluation with empirical speed data, and even fewer address the behaviour of heavy vehicles, whose suspension systems and axle configurations interact differently with vertical deflection elements.

This study addresses these gaps by combining a comprehensive inventory of all 2823 speed bumps in Zagreb, detailed geometric inspection at 208 locations, and UAV-based trajectory extraction at eight representative sites. This multi-layered methodology provides a more robust basis for evaluating speed-management performance and yields empirical insights relevant for urban traffic-safety planning and the standardization of vertical traffic-calming devices.

2. Literature Review

To enhance safety at specific locations within road networks, vehicle speeds must be effectively reduced. Traffic calming measures, as defined by the Institute of Transportation Engineers (1997), involve physical modifications to road infrastructure aimed at lowering traffic flow speeds and increasing driver awareness. Such measures generally result in reduced operating speeds and improved safety for all road users [5].

Traffic calming interventions are commonly divided into two categories [6,7]:

• Horizontal modifications, which force drivers to adjust their trajectory (e.g., road narrowing, chicanes, or roundabouts).
• Vertical modifications, which introduce abrupt changes in road surface elevation (e.g., speed bumps or raised platforms), compelling drivers to decelerate.

In Croatia, speed bumps are among the most widely applied traffic calming measures. As vertical modifications, they directly interact with vehicle suspension systems, encouraging drivers to reduce speed to maintain comfort and prevent mechanical damage at higher velocities [8].

Despite these advantages, many researchers have highlighted their drawbacks. For example, traversing a speed bump can increase travel time by 2.3 to 13.44 s per bump, depending on design and vehicle type [9]. This delay may translate into economic losses and hinder the timely arrival of emergency services. Frequent deceleration and acceleration cycles at speed bumps also elevate fuel consumption, which in turn worsens environmental impacts [6,10]. Studies indicate that fuel consumption can increase by up to 40% in networks with a density of 1.66 bumps per kilometer. Higher fuel use is directly associated with increased emissions of CO₂, CO, hydrocarbons (HC), nitrogen oxides (NOx), and particulate matter (PM) [11]. Moreover, braking before speed bumps generates additional particulate emissions due to friction and wear of brake pads and clutches [12].

These findings underscore the need for a balanced approach that maximizes safety benefits while minimizing adverse effects on traffic flow and the environment [6,13,14,15]. Recent research supports this dual perspective: Obregón-Biosca et al. [11] emphasized that improperly designed speed bumps can increase emissions, Kosakowska [7] demonstrated that raised platforms significantly reduce pedestrian accidents in Polish cities, and Džambas et al. [15] analyzed the trade-off between speed reduction and vehicle wear in several European urban contexts.

Despite these contributions, much of the existing research relies on relatively small or geographically narrow samples, often examining only a single bump type or a limited set of locations. Moreover, many studies do not distinguish between vehicle categories, although passenger cars, buses, and heavy trucks differ significantly in suspension characteristics, ground clearance, and speed adaptation behaviour. Only a few studies systematically assess whether installed bumps comply with national design standards, even though compliance can strongly influence driver response, comfort, and safety performance. These methodological limitations underscore the need for a broader, regulation-based evaluation that accounts for multiple bump types and vehicle categories under real-world traffic conditions—an approach adopted in the present study.

In addition to physical measures, recent research highlights the importance of modeling traffic network resilience in the context of system-wide disruptions, including cyber-physical risks. Liu et al. [16] proposed a comprehensive resilience framework that integrates failure-propagation modeling with machine learning-based forecasting. Although primarily oriented toward cyber-induced disruptions, the principles of proactive resilience planning are applicable to physical infrastructure interventions, including traffic calming strategies.

3. Materials and Methods

Ensuring safety is one of the fundamental objectives of transportation systems. Safety represents a key indicator of service quality, and related requirements should be considered during the planning and design phases of road infrastructure. Throughout this paper, the term speed bump is used as a general expression for vertical traffic calming measures; specific designs (e.g., raised platforms, rubber bumps, cushions) are specified where necessary.

3.1. Study Area and Data Collection

The analysis of speed bumps in the Republic of Croatia was conducted using the urban road network of the City of Zagreb, which contains a total of 2823 speed bumps. The primary objective of this research was to determine the extent to which speed bumps influence vehicle speed reduction and, consequently, improve traffic safety. To ensure reliability and validity, the study also assessed the compliance of installed speed bumps with applicable regulatory requirements and design standards.

All speed bump locations in Zagreb were first identified, after which a sample of 208 sites was selected for detailed investigation (Figure 1). The selection criteria required that each city district include at least 12 speed bumps, ensuring both geographical coverage and diversity in bump types. Representative distribution was prioritized to capture different designs and materials. Intersections and pedestrian crossings were not excluded, as the intent was to analyze driver behavior under realistic urban conditions in which these features frequently co-occur with traffic calming devices. Surrounding infrastructure elements, such as curves and crossings, were recorded at each site. While such features may influence driver behavior, they were considered inherent aspects of real-world urban environments and therefore included in the study.

Following site selection, each location underwent a comprehensive field inspection. This involved detailed measurements of both the surrounding roadway and the speed bump itself to evaluate compliance with regulatory design standards [17].

3.2. Measurement Procedure

Field research was carried out to assess compliance with national traffic regulations (NN 92/2019) [18]. The inspections included measurements of speed bump dimensions, surface material, and installation quality.

In addition, vehicle speeds were recorded at eight sites within the city, both prior to reaching the speed bump and directly on the bump itself. Speed measurement was conducted using unmanned aerial vehicles (UAVs) combined with DataFromSky software. This platform applies advanced computer vision and machine learning algorithms to process aerial footage and extract vehicle trajectories, speeds, and classifications [19,20].

UAV-based monitoring was selected because traditional roadside devices may influence driver behavior when visible, leading to artificially reduced speeds. In contrast, UAV observations remain unnoticed by drivers, ensuring that natural driving behavior is captured without external interference [20,21,22].

At each analyzed location, two measurement points were strategically positioned (Figure 2). The first point was located approximately 120–135 m upstream of the speed bump, while the second was placed directly on the bump. This setup enabled comparative analysis of vehicle speeds before and after traversal, providing insights into deceleration and acceleration patterns.

The monitoring process also included vehicle classification into five categories: passenger cars, heavy trucks, light trucks, buses, and motorcycles. UAV technology enabled precise tracking of these categories as they moved between measurement points. Vehicles observed at only one point were excluded from the dataset to ensure accuracy [23].

Figure 2 illustrates the placement of measurement points at one of the analyzed locations. In this example, vehicle speeds were recorded in the west–east direction. Vehicles detected at the first measurement point were tracked until they exited the zone at the second point, enabling a detailed analysis of their speed profiles.

At locations where vehicles were observed at only one measurement point, a single speed reading was obtained; such data were excluded from the analysis to preserve accuracy. UAV recordings were carried out at a fixed altitude of approximately 120 m, using 4K UHD video resolution (3840 × 2160 px, 29.97 fps, bit rate ~69,656 kbps), ensuring sufficient clarity for reliable vehicle detection and trajectory extraction. Speed data were obtained using DataFromSky software, which applies computer-vision-based object recognition and trajectory-tracking algorithms. Although detailed algorithmic parameters are proprietary, the software relies on an internally calibrated detection threshold and a temporal matching window adapted to urban environments with moderate traffic density.

Each site was recorded for approximately 15 min, corresponding to the maximum effective duration of a UAV battery cycle. All recordings were performed during daylight hours between 10:00 and 14:00 under clear, dry weather, good visibility, and stable traffic conditions. Periods of rain, fog, reduced visibility, and peak-hour congestion were avoided to minimize external influences on vehicle speed. Traffic volumes corresponded to typical mid-day conditions on residential and collector roads in Zagreb.

To ensure methodological consistency and enable replication, all environmental and traffic conditions were systematically documented for each UAV session. A summary of these parameters is presented in

Table 1. Summary of environmental and traffic conditions during UAV data collection.

Parameter	Condition	Description
Time of recording	10:00–14:00	Off-peak daytime hours, reduced congestion variability
Weather	Clear, dry	No rain, fog, or precipitation during recording
Visibility	>1 km	Stable visibility for UAV operations
Traffic density	Moderate/free-flow	Typical for residential and collector streets
Ambient temperature	15–25 °C	Stable meteorological conditions
Lighting conditions	Natural daylight	Consistent illumination across all recordings
Wind speed	<5 m/s	Within safe UAV operational limits

. The target reference speed for the safety assessment was 40 km/h, in accordance with national regulations for residential streets and school zones (NN 92/2019).

To ensure clarity and reproducibility, the main steps of the methodological approach are summarized as follows Figure 3.

The analysis relied exclusively on empirical field observations and standard statistical procedures. No AI-based predictive models were applied; vehicle trajectories were extracted using Data From Sky, which employs computer-vision-based detection and tracking. The key variables considered in the analysis included bump height, length, width, surface material, lane coverage, posted speed limits, and vehicle speeds at both measurement points. This structured approach provides a consistent basis for comparing different bump types and enhances the transparency of the overall analytical process.

3.3. Data Analysis

The compliance of each speed bump type with national regulations was assessed based on height, length, surface material, and proper signage. Field inspections in Zagreb identified five distinct types of speed bumps (Figure 4): raised platforms, rubber square bumps, pillow-type bumps, rubber pillow bumps, and narrow platforms.

Raised platforms represented the largest share of the analyzed sample (101 of 208) and demonstrated the highest compliance rate of 85% (

Table 2. Summary of speed bump compliance and analysis in the City of Zagreb.

Speed Bump Type	Number Analyzed	Compliance (%)
Raised platforms	101	85%
Rubber square bumps	52	60%
Pillow-type bumps	37	72%
Rubber pillow bumps	13	40%
Narrow platform	3	100%

). These structures are durable and generally meet design standards, making them effective in reducing vehicle speeds. Narrow platforms, although less common with only five cases, achieved full compliance (100%), reflecting high-quality design and implementation.

By contrast, rubber pillow bumps exhibited the lowest compliance rate (40%), largely due to material degradation and inconsistent installation practices. Rubber square bumps and pillow-type bumps showed moderate compliance rates of 60% and 72%, respectively, indicating variability in both design and maintenance.

Each type of speed bump was evaluated according to criteria such as height, length, surface material, and signage compliance. Raised platforms showed the highest compliance (85%) due to their robust construction and consistent adherence to design standards. Narrow platforms were the least represented category in the sample, with only three recorded locations. These three are also the only known remaining cases within the City of Zagreb. Consequently, they are being phased out from the road network and are not considered in future traffic calming projects. By contrast, rubber pillow bumps exhibited the lowest compliance (40%), largely due to material degradation, modular construction, and improper installation practices. Rubber square bumps (60%) and pillow-type bumps (72%) showed moderate compliance, reflecting variability in design and maintenance.

Overall, raised platforms appear to be the most suitable type of speed bump, as they span the full roadway width, are consistently marked with color, rarely suffer damage, and comply with all regulatory requirements. In contrast, rubber pillow bumps typically do not cover the entire roadway, often exceed the maximum allowed height, and are more prone to damage.

Table 3. Geometric measurements of each speed bump.

Speed Bump Type	Height (cm)	Width (cm)	Length (cm)
Raised platform	7.79	608	511
Rubber square bump	7.72	179	263
Pillow-type bump	7.05	206	373
Rubber pillow bump	9.22	201	277
Narrow platform	4.58	723	138

presents the results of on-site geometric measurements for each speed bump type. Only raised platforms fully satisfied all dimensional requirements prescribed by Croatian regulation NN 92/2019 [18].

4. Results

Based on the obtained results, a detailed analysis of all input parameters was conducted. A tabular format was used to display the number of vehicles and their average speeds across different categories within the measurement areas. Vehicle passages were also recorded according to their entry time at each measurement point. Figure 5 shows the spatial distribution of the eight speed bump sites in Zagreb, where UAV-based traffic counting and vehicle speed measurements were performed. Red markers indicate the locations of analyzed speed bumps.

In total, 906 vehicle passages were analyzed using UAV-based tracking. The sample included 862 passenger cars, 94 light trucks, 26 heavy trucks, 17 buses, and 7 motorcycles. Only vehicles with complete trajectories across both measurement points were included in the dataset to ensure data accuracy. It should be noted, however, that several vehicle categories, most notably heavy trucks, buses, and motorcycles, were represented by very small samples at certain locations. These low counts limit the statistical robustness of category-specific results, and the corresponding findings should therefore be interpreted with appropriate caution.

4.1. Traffic Counting and Speed Measurement in the Pillow Type Bump Zone

At this site, the distance between measurement points was approximately 120 m. The zone included two three-leg intersections, which influenced speed variability. A total of 125 vehicles were analyzed, with the highest recorded speed of 102.6 km/h. The average entry speed at Point 1 was 38.5 km/h, while the average speed on the bump (Point 2) was 29.7 km/h, representing a 23% reduction relative to the posted limit of 50 km/h.

Table 4. Average vehicle speeds in the pillow-type bump zone.

Vehicle Category	Average Speed (km/h)—Point 1	Average Speed (km/h)—Point 2	Change (%)
Passenger Car	38.90	29.68	−23.7
Heavy Truck	19.91	18.28	−8.19
Light Truck	42.02	33.56	−20.13
Bus	32.57	32.57	0
Motorcycle	/	37.13	/

presents the average speeds by vehicle category. Passenger cars and light trucks showed significant reductions (−23.7% and −20.1%, respectively), while heavy trucks experienced only minor reductions (−8.2%). Buses recorded no changes, which may be attributed to the presence of a bus stop in the measurement area. Motorcycles were excluded due to insufficient observations.

Data analysis indicates that a total of 125 vehicles passed through the first measurement point, with the highest recorded speed reaching 102.6 km/h. The average speed at this point was 38.5 km/h, corresponding to a 23% reduction relative to the posted limit of 50 km/h in residential areas. These results suggest that pillow-type bumps effectively reduce the speeds of passenger cars and light trucks, with reductions ranging from 8% to 24%. The presence of a bus stop within the measurement zone may explain the absence of speed reduction among buses.

4.2. Traffic Counting and Speed Measurement in the Narrow Platform Zone

At this site, the distance between measurement points was 125 m. The measurement area included access points to a shopping center and an elementary school courtyard, a signalized intersection at the entry, and a marked pedestrian crossing immediately before the bump. A total of 50 vehicles passed through Point 1, with a maximum recorded speed of 60.5 km/h and an average of 45.8 km/h. At Point 2, 62 vehicles were recorded, with a maximum of 32.8 km/h and an average of 20.9 km/h. This represents a 48% reduction relative to the posted speed limit of 40 km/h.

Table 5. Average vehicle speeds in the narrow platform zone.

Vehicle Category	Average Speed (km/h)—Point 1	Average Speed (km/h)—Point 2	Change (%)
Passenger Car	46.23	20.91	−54.77
Light Truck	35	20.24	−42.17

presents the average speeds by vehicle category. Passenger cars reduced speed from 46.23 km/h to 20.91 km/h (−54.8%), while light trucks reduced speed from 35.0 km/h to 20.24 km/h (−42.2%).

These results confirm that narrow platform bumps significantly reduce vehicle speeds. The substantial reductions across vehicle categories highlight their effectiveness, particularly in areas requiring strict speed control such as school zones and pedestrian crossings. Due to the low frequency of other vehicle types during the measurement period, only passenger cars and light trucks were included in the statistical analysis.

4.3. Traffic Counting and Speed Measurement in the Rubber Pillow Bump Zone

At this site, the distance between measurement points was approximately 125 m in the north–south direction. The zone included a three-leg intersection, a pedestrian crossing, and a nearby primary school, all of which influenced vehicle speed behavior.

Table 6. Average vehicle speeds in the rubber pillow bump zone.

Vehicle Category	Average Speed (km/h)—Point 1	Average Speed (km/h)—Point 2	Change (%)
Passenger Car	36.54	35.36	−3.23
Heavy Truck	23.19	26.48	14.19
Light Truck	49.77	48.30	−2.95
Bus	15.76	28.55	81.15
Motorcycle	44.33	43.10	−2.77

presents the average speeds by vehicle category. Passenger cars and light trucks showed negligible reductions (−3.2% and −3.0%, respectively), while motorcycles also recorded only a slight decrease (−2.8%). By contrast, heavy trucks and buses exhibited increases in speed (+14.2% and +81.2%), suggesting that larger vehicles were able to traverse the bump with minimal impact.

The results indicate that rubber pillow bumps do not significantly reduce vehicle speeds, except for passenger cars and light trucks. In contrast, heavy trucks and buses exhibited speed increases, likely due to their ability to traverse these bumps with minimal suspension impact.

4.4. Traffic Counting and Speed Measurement in the Rubber Square Bump Zone

Measurements were conducted at two sites to obtain representative data. The distance between measurement points was 130 m at the first location and 125 m at the second. Both areas included three-leg intersections, bus stops, and pedestrian crossings, all of which influenced traffic conditions.

Table 7. Average speeds of traffic entities in the rubber square speed bump zone.

Vehicle Category	Average Speed (km/h)—Point 1	Average Speed (km/h)—Point 2	Change (%)
Passenger Car	47.00	30.07	−36.02
Heavy Truck	31.87	34.25	7.47
Light Truck	46.46	36.60	−21.22
Bus	29.53	43.69	47.95
Motorcycle	43.30	37.37	−13.7

presents the average speeds by vehicle category. Passenger cars and motorcycles showed the most notable reductions (−36.0% and −13.7%, respectively). Light trucks also decreased moderately (−21.2%). In contrast, heavy trucks and buses recorded increases (+7.5% and +48.0%, respectively), suggesting that larger vehicles were able to traverse these bumps with minimal impact.

These results suggest that rubber square bumps are effective in reducing speeds for passenger cars and motorcycles but provide little to no reduction for heavy trucks and buses, which can traverse them with minimal impact.

4.5. Traffic Counting and Speed Measurement in the Raised Platform Zone

Measurements were conducted at three sites, with distances between measurement points ranging from 115 to 130 m. The average entry speed across these sites was 53.8 km/h, while the average speed on the platform was 26.1 km/h, representing a 35% reduction relative to the posted speed limit of 40 km/h.

Table 8. Average speeds of traffic entities in the raised platform speed bump zone.

Vehicle Category	Average Speed (km/h)—Point 1	Average Speed (km/h)—Point 2	Change (%)
Passenger Car	54.48	26.73	−50.94
Heavy Truck	51.11	22.22	−56.53
Light Truck	53.23	25.67	−51.78
Bus	46.29	18.57	−59.88
Motorcycle	37.25	20.80	−44.16

presents the average speeds by vehicle category. All categories showed substantial reductions, ranging from −44.2% (motorcycles) to −59.9% (buses).

Raised platforms proved to be the most effective traffic calming measure, reducing vehicle speeds by more than 50% across all categories. This substantial impact confirms their suitability as the optimal choice in areas where maintaining low vehicle speeds is critical, such as near schools and pedestrian crossings.

To provide a consolidated comparison across all bump types,

Table 9. Summary of percentage speed change by bump type and vehicle category.

Speed Bump Type	Passenger Car	Heavy Truck	Light Truck	Motorcycle	Bus	Overall (Avg.)
Pillow-type bump	−23.70%	−8.19%	−20.13%	—	0%	−13.00%
Narrow platform	−54.77%	—	−42.17%	—	—	−48.47%
Rubber pillow bump	−3.23%	14.19%	−2.95%	−2.77%	81.15%	17.28%
Rubber square bump	−36.02%	7.47%	−21.22%	−13.70%	47.95%	−3.10%
Raised platform	−50.94%	−56.53%	−51.78%	−44.16%	−59.88%	−52.66%

summarizes the exact percentage speed changes for all vehicle categories without rounding.

Several measurement sites included contextual elements such as nearby schools, pedestrian crossings, signalized intersections, bus stops, or commercial access points. Although these features were not incorporated as separate quantitative variables in the statistical analysis, their influence on driver behavior was clearly observed during fieldwork. Locations near schools or pedestrian crossings generally exhibited lower approach speeds, while wider mid-block sections or areas without pedestrian activity tended to produce higher entry speeds and smaller relative reductions. These contextual effects help explain some local variations in speed profiles but do not affect the relative ranking of the speed bump types, which remained consistent across all observed sites. A more detailed quantification of these environmental factors is recommended for future research.

To complement these descriptive results, a statistical comparison of speed reductions across bump types was also performed. An ANOVA test, including variance measures and a boxplot illustrating distribution differences, is presented and interpreted in the Discussion section (Section 5). In addition, standard deviations and variances were calculated for all bump types to quantify variability, and the distribution of speed reductions is shown using boxplots.

5. Discussion

A detailed analysis of traffic counts and vehicle speed measurements was conducted across five types of speed bumps at eight locations in Zagreb.

Table 10. Average vehicle speeds by category on different types of speed bumps.

Type of Speed Bump	Avg. Speed on Bump (km/h)	Average Speed by Vehicle Category (km/h)
		Passenger Car	Heavy Truck	Light Truck	Motorcycle	Bus
Pillow-type bump	29.67	29.68	18.28	33.56	32.57	37.13
Narrow Platform	20.88	20.91	-	20.24	-	-
Rubber Pillow bump	35.74	35.36	26.48	48.30	28.55	43.10
Rubber Square bump	30.81	30.07	34.25	36.60	43.69	37.37
Raised Platform	26.08	26.73	22.22	25.67	20.80	18.57

summarizes the average vehicle speeds observed for each bump type, including values disaggregated by vehicle category.

The highest average speed was observed on rubber pillow bumps (35.7 km/h), followed by rubber square bumps (30.8 km/h). The lowest speeds occurred on narrow platforms (20.9 km/h) and raised platforms (26.1 km/h). When analyzed by vehicle category, light trucks reached the highest residual speeds (48.3 km/h on rubber pillow bumps), while heavy trucks recorded the lowest (18.3 km/h on pillow-type bumps). Across all devices, average vehicle speeds on bumps were lower than the posted limit of 40 km/h. In percentage terms, narrow platforms showed the greatest deviation from the limit (−48%), while rubber pillow bumps achieved the smallest (−11%). These findings confirm that while all bump types reduce speed, their effectiveness varies substantially. Devices that consistently lowered speeds below the 40 km/h threshold, particularly raised platforms, demonstrated the highest safety performance.

Several limitations should be considered. First, the study covered eight locations, which may not fully represent the range of urban traffic environments. Road conditions, traffic density, and driver behavior may have influenced recorded speeds, limiting generalizability. Second, reductions expressed only in percentage terms may not fully reflect safety performance; therefore, effectiveness was also assessed relative to the 40 km/h target speed. Third, external factors such as weather, time of day, and vehicle load conditions were not controlled. For instance, speed profiles of heavy trucks and buses can vary depending on whether they are fully loaded or empty. Finally, data collection was based on snapshot measurements rather than longitudinal monitoring, which could provide deeper insight into how drivers adapt to bumps over time. Compliance with posted limits prior to reaching the bump was also not explicitly considered. Beyond physical and environmental factors, broader macroeconomic influences may also shape traffic safety outcomes. Recent research demonstrated that fluctuations in fuel prices can significantly affect accident rates across Europe [24].

Despite these limitations, the findings provide a robust foundation for understanding the impact of different bump types on speed reduction. Rubber pillow bumps permitted the highest residual speeds (up to 48 km/h for light trucks), whereas raised platforms consistently reduced speeds to below 27 km/h across all categories. These results align with Kosakowska [6], who reported that platform-type bumps ensure better compliance in Polish urban areas, and with Obregón-Biosca [10], who found that the modular rubber pillow bump is often ineffective for heavy vehicles. A notable and somewhat counterintuitive finding emerged at locations with modular rubber speed bumps, where heavy trucks and buses occasionally recorded higher speeds instead of slowing down. This behavior can be explained by a combination of the bumps’ geometry and the structural characteristics of large vehicles. Modular rubber units have a width of approximately 201 cm, which is significantly narrower than the wheel track of heavy vehicles. Because they do not cover the full lane width, truck and bus drivers can often position the vehicle so that one side partially or completely bypasses the vertical deflection, reducing the effective impact that the bump is intended to generate.

In addition, the measured height of 9.22 cm affects different vehicle categories unevenly. For passenger cars, this height causes noticeable suspension compression and naturally enforces speed reduction. For heavy vehicles, however, the suspension is designed for substantially higher loads, and its rigidity means that a relatively short and narrow rubber element often does not produce meaningful vertical displacement. Consequently, both the mechanical impact and the driver’s perceived need to decelerate are diminished. When combined with the possibility of slight lateral maneuvering to avoid the bump’s center, this can even result in higher traversal speeds than expected.

These results indicate that modular rubber speed bumps are not a suitable choice in areas with a high share of heavy truck or bus traffic. Their effectiveness under such conditions is limited, and their use should be reconsidered in favor of wider and structurally more robust vertical traffic-calming devices.

Furthermore, as over 45% of Zagreb’s speed bumps do not meet national regulatory standards (Section 3.3), the presented framework offers an important tool for future standardization and safety planning.

The dimensions and application of raised platforms in Zagreb closely align with the Dutch CROW Guidelines [25], widely recognized across European cities as best practice in traffic calming. Although this study was based on Croatian regulation NN 92/2019, its findings are relevant for broader European contexts. From a safety perspective, bumps that reduce average speeds to near or below the 40 km/h target threshold offer greater protection to vulnerable road users. Raised platforms and narrow tables achieved the strongest results, while rubber pillow bumps frequently allowed speeds above the threshold, limiting their effectiveness, particularly in pedestrian- and cyclist-rich environments.

To examine statistical differences between the five speed bump types, descriptive statistics for each group (sample size, mean values, and variance) were first calculated, as shown in

Table 11. Descriptive statistics of speed reduction across different speed bump types.

Groups	Count	Sum	Average	Variance
Rubber pillow bump	45	8.99	0.20	102.99
Pillow-type bumps	103	976.28	9.48	76.18
Narrow platform	47	1060.30	22.56	190.56
Rubber square bump	115	2170.71	18.88	239.13
Raised platform	154	4214.81	27.37	90.98

. These values served as the input dataset for the subsequent analysis of variance (ANOVA). An analysis of variance (ANOVA) was performed to determine whether statistically significant differences exist in speed reduction among five types of speed bumps (rubber pillow bumps, pillow-type bumps, narrow platforms, rubber square bumps, and raised platforms). The results revealed a highly significant F-statistic (F = 66.93) accompanied by an extremely low p-value (p < 0.001), clearly indicating that the observed differences in speed reduction between groups are statistically significant (

Table 12. One-way ANOVA results for speed reduction by bump type.

Source of Variation	SS	df	MS	F	p-Value	F Crit
Between Groups	36,307.06	4	9076.76	66.93	0.001	2.39
Within Groups	62,248.69	459	135.62
Total	98,555.75	463

). Since the empirically obtained F-value substantially exceeds the critical threshold (F_crit = 2.39), it can be concluded that bump type exerts a significant and measurable influence on vehicle speed reduction.

The results suggest that different bump types vary considerably in their effectiveness. The box-plot (Figure 6) illustrates the distribution of speed reduction by bump type, showing the median, interquartile range (IQR), minimum and maximum values (within 1.5 × IQR), and potential outliers. Rubber pillow bumps exhibited the least reduction, while raised platforms produced the largest effect with the greatest variability. These findings reinforce the ANOVA results, confirming statistically significant differences among bump types (F = 66.93, p < 0.001).

These findings are consistent with another study [26], which showed that demographic factors such as age and gender significantly influence accident causation, with younger and older drivers particularly prone to excessive speed. This reinforces the importance of adapting traffic calming devices to the demographic structure of road users. From a practical perspective, these findings suggest that municipalities should prioritize the installation of raised platforms and narrow platforms in areas requiring strict speed control, such as school zones and residential neighborhoods. In contrast, the continued use of modular rubber pillow bumps should be reconsidered, as their limited effectiveness in reducing speeds—particularly for heavy vehicles—may undermine the overall efficiency of traffic calming strategies. These findings are further supported by Antić et al. [4], who showed that higher bumps (5 and 7 cm) in Belgrade resulted in substantially greater speed reductions compared to lower bumps (3 cm), confirming that both bump geometry and height play a critical role in ensuring effective speed control. The observed variations in speed reduction also align with the congestion boundary approach [27], which interprets localized speed changes as dynamic transitions between free-flow and congested states. Similarly, Laval [28] conceptualized traffic flow as a physical fluid, demonstrating that congestion behaves like a phase transition between gas-like (free-flow) and liquid-like (jammed) states. These perspectives provide a broader theoretical framework for understanding how local interventions, such as speed bumps, can trigger short-term phase shifts in urban traffic dynamics.

In addition, the consistency of the findings across eight UAV-observed sites supports the robustness of the applied methodology. Despite variations in surrounding road geometry, traffic composition, and local driving behavior, the relative performance ranking of bump types remained stable, strengthening the validity of the comparative approach. A formal sensitivity analysis was not conducted, as the study is primarily empirical; however, the multi-site design and cross-comparison between bump categories reduce the likelihood that the observed effects are location-specific.

From a planning perspective, the results have several practical implications. Locations with frequent pedestrian activity—such as school zones, residential streets, and areas with vulnerable road users—should prioritize the installation of raised platforms or narrow tables, as these devices consistently enforce speeds well below the 40 km/h safety threshold. Conversely, modular rubber bumps, especially pillow-type designs, should be avoided on streets with a high proportion of heavy trucks or buses, as their geometry allows partial bypassing and results in limited speed control. For city administrations, the findings also highlight the importance of systematic compliance checks, given that almost half of the existing devices in Zagreb do not meet national standards. Establishing uniform design guidelines and replacing underperforming bump types could substantially improve the reliability of speed management across the network. Overall, the study provides evidence-based guidance that can support municipalities in selecting appropriate traffic-calming measures and planning future interventions more effectively.

6. Conclusions

This study examined the effectiveness of five types of speed bumps installed across the urban road network of Zagreb, using a combined approach that included geometric assessment, UAV-based speed measurements, and statistical analysis. The results showed clear differences in performance between the analyzed devices. Raised platforms proved to be the most effective, consistently reducing speeds to well below the posted 40 km/h limit, while narrow platforms also achieved notable reductions. In contrast, modular rubber devices—especially rubber pillow bumps—demonstrated the weakest performance and often allowed residual speeds above the desired threshold.

Differences between vehicle categories further highlight the importance of selecting appropriate devices. Light trucks and buses tended to maintain higher speeds over modular bumps, while the suspension characteristics of heavy vehicles reduced the impact of narrow or low-profile designs. ANOVA results confirmed that the differences between bump types were statistically significant, emphasizing the need to align device selection with road function and traffic composition.

Several limitations should be acknowledged. Measurements were carried out at a limited number of sites and within short observation periods, and external factors such as weather conditions, vehicle loading, and long-term driver adaptation were not systematically assessed. Future studies should include longitudinal monitoring, comparisons of different geometric variants, and evaluations of secondary effects such as fuel consumption and emissions.

From a practical standpoint, the findings suggest that wider, full-width vertical measures—particularly raised platforms—should be prioritized in areas where strict speed control is required. Conversely, the widespread use of modular rubber bumps should be reconsidered in locations with a higher share of heavy-vehicle traffic, given their limited effectiveness.

Although this research was conducted in Zagreb, the applied methodological framework—combining UAV-based monitoring with compliance assessment—can be replicated in other cities aiming to evaluate and optimize traffic-calming strategies.

The results contribute to a more evidence-based selection of speed management measures in urban environments and support the broader development of standardized traffic-calming practices.