do speed bumps really decrease traffic speed? an italian experience

Accident Analysis and Prevention 33 (2001) 585–597

Do speed bumps really decrease traffic speed?An Italian experience

Massimiliano Pau *, Silvano AngiusDepartment of Territorial Engineering, Transportation Section, Uni�ersity of Cagliari, Piazza d’Armi, 09123 Cagliari, Italy

Received 2 March 2000; received in revised form 30 July 2000; accepted 31 July 2000

Abstract

Italy introduced the extensive use of speed bumps only in 1990, in an attempt to limit the high number of fatalities involvingpedestrians in urban streets caused by the high speed of vehicles. In many countries, such devices have been the subject of carefulinvestigations (in order to assess their effectiveness and disadvantages for the traffic circulation) and this has resulted in a numberof modifications in the design to improve their performance. On the contrary, no systematic and scientific studies have beencarried out on Italian installations: moreover, the type of undulation adopted is known to produce a series of problems for somecategories of users and is not so effective in reducing speed as larger devices such as ‘speed humps’ or ‘speed cushions’. This paperproposes a study of the effectiveness of 23 speed bumps installed in the city of Cagliari; to this aim, a speed analysis wasperformed at speed bump locations, at the crosswalks protected by the devices and at sections of the streets where bumps areinstalled but far from them. The results show that in one third of the cases the 85th percentile of speed measured at the speedbumps is higher than the posted speed limit (50 km/h) and an equal percentage of vehicles travel at a speed in the range of 45–50km/h. No statistically significant differences were found from the comparison of speed values observed in free, bump or crosswalksections of the same streets, while speed profiles calculated at four sites, where a high percentage of braking vehicles was observed,showed a common trend from which it clearly emerges that the effect of the device on driver’s behaviour is restricted to a shortspatial range (about 20–30 m before and after the bump). The current situation thus suggests the use of more effective devicessuch as humps or cushions, or the integration of speed bumps with other traffic calming techniques. © 2001 Elsevier Science Ltd.All rights reserved.

Keywords: Speed bumps; Pedestrian safety; Traffic calming

www.elsevier.com/locate/aap

1. Introduction, background and purposes of the study

Since early studies were carried out by the Trans-portation Research Laboratory (UK) in the 1970s(Watts, 1973), the use of vertical raisings of the roadpavement as a passive method for controlling the speedof vehicles has become common in many countries(especially in western Europe and North America).

To date, several kinds of these devices have beendesigned by exploiting the same basic principles, butwith the obtaining of different results in terms of effec-tiveness, discomfort, and dangerousness for some cate-gories of road users; usually they are classified as:

1. Speed ‘‘bumps’’.2. Speed ‘‘humps’’.3. Speed ‘‘cushions’’.

The first two are characterised by a continuous verti-cal deflection placed across the street made using sev-eral materials (rubber, concrete, thermoplastic materialsetc.) but with a range of different dimensions. Asshown in Fig. 1, speed bumps are very short (usuallythe width is between 600 and 1200 mm) with a height inthe range of 30–100 mm (depending on the postedspeed limit) and a circular or parabolic profile, whilespeed humps have about the same height values, butare wider (from 4000 to 8000 mm) and have a variety ofdifferent profiles (circular, sinusoidal, flat top etc.).

From the viewpoint of the effect on vehicles, speedbumps produce impacts that are often dangerous forthe suspension system and are a source of high levels of

* Corresponding author. Tel.: +39-070-6755270; fax: +39-070-6755261.

E-mail address: [email protected] (M. Pau).

0001-4575/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.PII: S0001-4575(00)00070-1

M. Pau, S. Angius / Accident Analysis and Pre�ention 33 (2001) 585–597586

discomfort, especially for some categories of road users,such as cyclists, motorcyclists and occupants of emer-gency vehicles.

Moreover, many drivers have found that an increasein speed tends (within a certain range of speed) toreduce the magnitude of the vertical acceleration, con-sequently forcing the vehicle to increase its speed toattenuate the jolt and thus producing a significantreduction of the effectiveness of such a device (Watts,1973; Kassem and Al-Nassar, 1982).

On the other hand, the lower ramp of the speedhumps and their special profiles allow the vehicle to

Table 1Range of speed observed in past studies where various types ofundulations are employed

Range of speedDeviceInvestigatorobserved (85%) (km/h)

35.2–43.7Sumner and Baguleya (UK, Humps1979)

Humps 30.4–38.4Clementb (USA, 1983)22.4–32.6HumpsMakc (USA, 1986)20.8–41.6Stephensd (USA, 1986) Humps

Broadbent and Salmone 39.2–46.4Bumps(UK, 1991)

18–20HumpsEngel and Thomsenf

(Denmark, 1992)HumpsVis and othersg (The 25m

Netherlands, 1992)Humps 32–36Toronto City Councilh

(Canada, 1997)Webster and Layfieldi (UK, 32–45.3Humps

1998)Cushions 35.7mWebster and Layfieldj (UK,

1998)Layfield and Parryk (UK, Cushions 24.8–50.9

1998)Humps 42.9mBureau of Traffic

Management City ofPortlandl, (USA, 1998)

a Not specified method at various sites (UK).b Not specified method at Thousand Oaks (CL, USA).c Time-lapse video photography at San Antonio (TX, USA).d Synthesis of 18 studies performed in USA, UK and Australia.e Not specified method at Wakefield (UK).f Radar measures at Kogens Enge (Danmark).g Not specified method at various sites (The Netherlands).h Not specified method at Toronto (Canada).i Radar measures at Glenrothes (UK).j Radar measures at Northampton (UK).k Radar measures at various sites (UK).l Not specified method at Portland (OR, USA).m Mean value.

Fig. 1. Differences between various kind of undulations: From left toright (a) speed bumps, (b) speed humps, (c) speed cushions.

gently straddle the undulation, at the same time forcingthe driver to reduce the speed to suitable values inorder to avoid a catastrophic jump over the roadpavement.

The analysis of drivers’ response to installation ofsuch raisings shows that users prefer larger humps sincethey are more comfortable in terms of jolts perceivedinside the car, (Watts, 1973; Hodge, 1993; Webster andLayfield, 1998) a fact that is also confirmed by thevalues of internal vertical acceleration.

Previous tests focusing on the assessment of undula-tion effectiveness in terms of speed reduction alwaysshowed a decrease in the 85th and 50th percentile ofspeed (Sumner and Baguley, 1979; Mak, 1986; Vis etal., 1992; Layfield and Parry, 1998; Webster, 1998;Webster and Layfield 1998) even though there weremarked differences depending on the device employed;a list of 85th percentile speed values recorded withvarious kinds of undulations are reported in Table 1.

M. Pau, S. Angius / Accident Analysis and Pre�ention 33 (2001) 585–597 587

As regards speed cushions (introduced in Germany in1992) they basically consist of a single raised area,usually smaller than lane width, with a squared orrectangular shape, especially designed to avoid exces-sive discomfort (or damage) to large trucks, buses andemergency vehicles.

While such devices allow large vehicles to straddlethem practically without any effect, smaller vehicles areforced, due to their reduced track width, to ride overthe cushion with at least one couple of wheels. Al-though not so effective as large speed humps, thereduction in the 85th speed observed in past studies onspeed cushion schemes lies in the range of 27–35 km/h(Layfield and Parry, 1998).

1.1. The Italian situation

In Italy, the extensive use of speed bumps started in1993, as primary countermeasures to limit the largenumber of urban accidents with fatalities involvingpedestrians (more than 16000 accidents with about 700fatalities in 1998).

Of all possible traffic calming methods, the installa-tion of speed bumps was chosen mainly for their lowcost (about 700 Euro for a two 3.50 m lane street) andfor their easy installation, associated with reduced timeduring which the normal traffic circulation must bestopped or diverted to complete the works.

From the legal point of view, the ‘Italian TrafficCode’ (Codice della Strada, CdS) states that, in general,every speed reducing device must be installed only usingmodular elements made of rubber or plastic material,while the use of concrete is restricted to streets wherethe posted speed limit is 30 km/h.

The limitations on width and height of the devicesare expressed as follows.1. For streets where the posted speed limit is less than

or equal to 50 km/h, width must be greater than orequal to 600 mm and height greater than or equal to30 mm.

2. For streets where the posted speed limit is less thanor equal to 40 km/h, width must be greater than orequal to 900 mm and height greater than or equal to50 mm.

3. For streets where the posted speed limit is less thanor equal to 30 km/h, width must be greater than orequal to 1200 mm and height greater than or equalto 70 mm.

As regards the category of streets where speed reduc-ers could be installed, the traffic code states that onlyresidential streets, public and private parks and resi-dences can be sites suitable for this sort of trafficcalming technique: moreover, the placement of speedbumps on streets that are part of preferential routes foremergency vehicles is expressly forbidden.

Since speed bumps are ‘commonly’ perceived as ef-fective in reducing traffic speed, the local authoritiesemploy them frequently even without a systematic anal-ysis of the actual advantages and disadvantages, thusmaking the ‘use’ of such devices a ‘misuse’, as if it couldplay the role of a panacea for all kinds of speed-relatedproblems.

All the above-mentioned considerations explain thelack of scientific tests on the effectiveness of speedbumps (i.e. no regular ‘before and after’ studies arecarried out to assess the reduction of vehicle speeds),therefore, it is difficult to estimate if benefits from theinstallation of speed bumps (generalised speed reduc-tion, decrease in accidents involving pedestrians etc.)are greater than costs in terms of car damage, reductionin safety for some road users (motorcyclists and cy-clists) and increase in car accidents due to drivers’incorrect behaviour, such as sudden braking (or acceler-ating) and diversion of the trajectories to avoid crossingthe speed bump.

As the main reason for which the Italian authoritiesinstall speed bumps is the protection of pedestriancrosswalks, the general aim of this study is to assesswhether traffic speed is currently at levels compatiblewith the safety needs of weak traffic components. Toachieve this purpose, the entire system of speed bumpsin a middle-size town were examined in order to evalu-ate traffic speed characteristics both at the speed bumplocations, at crosswalks protected by them, and atsections far from the devices.

2. Materials and methods

2.1. Speed bumps

Speed observations were made in the city of Cagliari(180000 inhabitants) a middle-size town located in Sar-dinia (Italy’s second largest island).

The Traffic Bureau of Cagliari decided to installspeed bumps in 1996 ‘experimentally’ and, to date,there are 30 devices placed on streets with very differentgeometrical and functional characteristics, but with thecommon feature of a posted speed limit of 50 km/h.Seven of these locations were excluded from the study,since factors other than the presence of the undulationmay influence speed values (e.g. a close intersection, aturn, etc.). The streets were identified simply using thecriterion of assigning them a letter and a number thatrepresents the number of lanes. In this way, site ‘X1’classifies street X having one lane per direction of travel(see Table 2 for details).

Tested speed bumps are produced by 3M® Italy(Italian branch of 3M® Corp., USA), as single modulesof black vulcanised rubber covered with stripes of ayellow high-reflective tape (see Fig. 2) which must be


clamped together to compose the entire undulation:placement on the road surface is made using steel nails.The modules are 30 mm high and 600 mm wide with acircular profile and are suitable for streets where theposted speed limit is 50 km/h. The distance between

speed bumps and crosswalks to be protected is in allcases in the 20–25 m range. A typical section of aselected two lane street is illustrated in Fig. 3.

It is to be noted that this sort of undulation requirescareful maintenance, since local deformations and rup-

Table 2Geometrical features, flows and percentage of braking vehicles in the streets where speed bumps are currently installed

Total lane width per travel direction (m) Flowa (vehicles per hour)Site Percentage of braking vehiclesa (%)Number of lanes

5.95A1 14681 896403.15B1 1

C1 1 3.15 677 26D1 826931

2583 131E1F1 1 2.6 182 20G1 1 3.8 231 20

20H1 1 2364.5I1 354 1041

3.45 739J1 914.5K1 11601 9

2 5.8 1054 27bL213546 82M2

28b2N2 2827O2 990 31b7.22

6.8 909P2 1126.2Q2 16342 26

R2 2 5.4 1507 24S2 2 7.1 878 38b

235324T1 1U2 7.752 1320 15

3 8.7V3 1705 17215027.6W2 2

a Mean value of four observations.b Sites selected for the speed profile analysis.

Fig. 2. Rubber module used to compose speed bumps (courtesy 3M® Italy).


Fig. 3. Example of urban two-lane, two-way street where speed bumps are currently installed.

tures often occur, especially due to heavy traffic, thuscausing dangerous situations and, above all, potentialreductions in the effectiveness of the device.

2.2. Speed data acquisition

The acquisition of speed data was carried out bymeans of a Laser Traffic Counter (Sodi KV-Laser, SodiCo., Italy) properly placed on a car window (Fig. 4)parked parallel to the direction of traffic, in such a wayas to hide the presence of the instrument and to avoidpossible alterations of drivers’ natural behaviour (i.e.reductions of speed) which often occur when speed

acquisition devices (such as pneumatic tubes or radarplaced on a tripod, etc.) are clearly visible at the side ofthe street or on it.

The type of counter employed records speed andlength of vehicles, travel direction (since measurementscan also be performed in the opposite direction) andtotal traffic flow. The counter can also be connected toa PC to download all data via serial interface and adedicated software can perform basic processing, sup-plying the number of vehicles for each class of speed(from the interval 0–10 up to 240–250 km/h) and thepercentage of heavy vehicles.

In the first step of the study, data were collected:


1. at speed bump locations;2. at crosswalks protected by speed bumps;3. in streets where speed bumps are present, but at a

distance such that their influence on driver’s be-haviour can be considered exhausted (i.e. more than100 m before or after) (Mak, 1986; City of Portlandand Bureau of Traffic Management, 1998);

4. in streets where speed bumps are not installed.Each analysis was repeated four times, on an inter-

mediate weekday (Tuesday, Wednesday or Thursday)choosing off-peak hours, so as to perform the analysisin conditions as close as possible to the free flow state.This situation represents the worst condition for pedes-trian safety, since drivers are not conditioned by thepresence of close vehicles and tend to increase theirspeed up to values influenced only by the geometricalcharacteristics of the street. The time of acquisition was1 h for each site; thus a total of 92 h of samplingrepresents the data base for the study.

At the same time as the speed counter was recording,an operator annotated the number of vehicles thatbraked owing to the presence of the speed bumps: thiskind of behaviour is quite easy to verify, since it ishighlighted by the rear stop lights turning on. Thenumber of braking vehicles for each site is reported inTable 2.

After this first phase of the study, data were elabo-rated to identify the locations in which speed bumpsseemed to modify drivers’ behaviour significantly.

On the basis of the results of this process, four siteswere selected to perform the second step of the analysis,which consisted of a detailed study of the speed profilebefore and after the speed bump position: this investi-gation was carried out by collecting speed data from 13points in a 200 m interval centred on the undulationaxis.

In this case, in order to perform the data recordingunder similar traffic conditions, acquisition time wasreduced to 30 min for each point.

Finally, the relationship between traffic speed andtotal width of the lanes of the selected streets wasexamined, both for speed bump and free sections. Thepurpose of this analysis was to assess possible differ-ences between the two cases recognisable as introducedby the presence of the undulations.

To see if the speed variations observed during thecomparisons are statistically significant, one way analy-sis of variance (ANOVA) tests were performed and thecorrespondent P value calculated.

3. Results

3.1. Speed �alues at bumps, at related crosswalks andat ‘free’ sections

The overall results of the speed analysis performed inthe present study are outlined in the diagrams in thefigures Figs. 5–8, while the number of samples for eachanalysis or comparison, as well as the mean value of thespeed parameters and the ANOVA results, are given inTable 3.

Fig. 5 shows the recorded values of both the 85thand 50th percentile of speed for all sites with testedspeed bumps compared with the posted speed limit (50km/h).

Speeds can be ideally subdivided in three groups: inabout 30% of the cases, the 85th percentile speed ishigher than the posted speed limit, while at 26% of thesites the observed speed values lie in the 45–50 km/hrange; the remaining third of the tested speed bumpsreveal a speed under 45 km/h. As regards the meanspeed, more than 50% of the observations are in the30–40 km/h range while the remaining half of thevalues are almost equally divided between values higherthan 40 km/h (26%) and lower than 30 km/h (17%).

The subsequent analysis carried out in the study, andillustrated in the diagram in Fig. 6 aimed at seeing

Fig. 4. The car equipped with the laser traffic counter.


Fig. 5. 85th and 50th percentile speed recorded at the 23 speed bump installations.

whether a low speed value was assured at the cross-walks supposedly protected by the presence of thespeed bump. Thus, the comparison of speeds recordedat speed bumps and crosswalks immediately after themfor a sample of 11 streets out of a total of 23 isproposed, in this case more than 50% of the tested sitesshowed an increase in the 85th percentile speed beforethe speed bump location while in the remaining part ofthe sample; reductions in the range of 10–25% werefound. Similar results were observed for the 50th per-centile values, with a slight reduction in the number ofsites at which the increase in speed in the space betweenbump and crosswalk is present (45%). However, in boththe cases, the statistical analysis revealed that in thespace between the bumps and the crosswalks (20–25m), there was no significant change in speed.

The last test was performed by comparing data ob-tained both at the speed bump locations and at sectionsof the same street far from the undulation (where, aspreviously said, the term ‘far’ means a distance of atleast 100 m from the device). In this case the number ofsampled sites was reduced to eight, since a large dis-tance of both sections from any sort of influencingfactors (such as intersections, turns, parks etc.) wasrequired and it was impossible to assure such condi-tions at all the original 23 locations.

The results, reported in Fig. 7, show that in two casesthere are no differences in the speed induced by thepresence of the speed bump for the 85th percentile,while at most sites slight reductions in speed (in the1–18% range) were observed; a slight increase in speedat the bump location (probably due to random factors)was found only in section R2. On the contrary, all the50th percentile bump values except R2 were lower thanin the no bump situation. However, the statistical anal-ysis again revealed that for both percentiles, the speedvalues were not significantly affected by the presence ofthe bumps.

3.2. Speed profiles

On the basis of visual observations, four sites wereselected (with the criterion previously described) bysupposing that a higher percentages of braking vehiclesrepresented a situation in which drivers were moreinfluenced by the presence of the bumps. At theselocations, the analysis focused on calculating the speedprofiles by moving the laser counter in a range of 200m, centred on the speed bump axis, and recording thespeed values at 13 points (−100, −50, −40, −30 mand so on up to 100 m after the device), thus obtaining


the profile of 85th percentile speed along a possiblezone of influence.

Although literature in this field is scanty, previousstudies (Mak, 1986; City of Portland and Bureau ofTraffic Management, 1998; Barbosa et al., 2000) haveshown that speed should decrease and reach its mini-mum value at the device location, then increase tooriginal values. The spatial range in which the speeddecreases represents the influence zone of the undula-tion; the longer this region is the more effective thedevice is.

As shown in Fig. 8, the experimental points werefitted with a Gaussian function

y=a+b exp(− (x−b)/c)2)

which seemed to be suitable for all four cases with areasonable agreement.

On the basis of the above cited references, worthy ofnote is a common trend characterised by a decrease inspeed values in correspondence to the speed bumpposition or in its vicinity, while once the vehicles hadcrossed the undulation, drivers tended to increase their

Fig. 6. 85th and 50th percentile speed recorded at speed bump locations and at the crosswalks protected by them.


Fig. 7. 85th and 50th percentile speed recorded at speed bump locations and at sections of the same street at least 100 m from them.

speed quickly up to initial levels within a short distance.This observation confirms the existence of an influenceregion ranging from 30 to 60 m.

On the other hand, the amount of speed re-duction achieved, calculated on the basis of thehighest and lowest speeds, is not very high (10–17%);moreover, as partial confirmation of the previoustest, speed values recorded at the bumps were stillhigher than the posted speed limit in three of the foursites.

3.3. Relationship between speed and lane width

The data obtained from the speed bump locationanalysis suggested that speed values were more influ-enced by the geometrical features of the street ratherthan the presence of the traffic calming devices. Inorder to better assess this effect, the values of 85thpercentile speed measured over a certain section werecoupled with the street width value; the results of thiscomparison are reported in Fig. 9.


As can be seen, and as was reasonably expected, bothin the case of ‘free’ sections and speed bump sectionsspeed tended to increase as available lane width in-creased. These results are in agreement with previousstudies showing a positive relationship between speedand road width; in fact, roads characterised by excellentgeometrical features lead drivers to perceive high levelsof safety and underestimate travel speed (Fildes andLee, 1993).

A linear regression on the experimental points, per-formed in order to express an analytical relationshipbetween 85th speed and total available lane width, gavethe following results,

S=28.93+3.57 LW (bump case, R2=0.50)

S=19.19+6.29 LW (bump case, R2=0.49)

where S is the 85th percentile speed and LW is the totallane width (obtained by multiplying the number oflanes by the lane width).

Comparing the bump and the free situation, a veryslight difference can be observed, for example, in thecase of a typical urban street (two lanes of 3.50 m each)the expected decrease in speed induced by the presenceof a speed bump is about 10 km/h (from 63 to 53km/h), a value that can be considered quite low for thisclass of speed. The ANOVA, performed after the calcu-

Fig. 8. Speed profiles recorded in four streets in a 200 m range (100 m before and 100 m after the speed bump position).

Table 3Mean values of 85th and 50th percentile speeds and results of the statistical test for the comparison proposed in the study

Number of streets sampledType of section 85th speeda 50th speeda ANOVA P-statistic

Comparison 1(Fig. 7)Bumps –23 46.45 35.53

Comparison 2 (Fig. 8)38.6011 0.4769 (85th)50.87Bumps

0.5230 (50th)11 36.93Crosswalks protected by bumps 48.23

Comparison 3 (Fig. 9)57.168 0.3954 (85th)Bumps 45.65

0.1774 (50th)40.0752.538No bumps

a Mean value.


Fig. 9. Speed-lane width relationship both for speed bump installations and free sections.

lation of speed values on the basis of the two regressionlines (for lane widths in the range of 3–10 m) shows nosignificance (P=0.12), thus confirming, even in thiscase, that there is no statistical evidence supporting thegeneral conviction that speed bumps are really effectiveas speed reducing devices.

4. Discussion

Do really narrow speed bumps reduce traffic speed?Although we are aware that further data must becollected in order to provide a scientifically based an-swer, the first impression on the effectiveness of thesedevices is not positive.

As shown by the results, the main aspect to highlightis that the observed speed values at device locations aresignificantly higher than any other value recorded insimilar surveys in the past. We are not able to statewhether a real benefit was obtained from the installa-tion of speed bumps; it is possible that the past situa-tion was even worse than the present (traffic in Italiancities is well known to be rather chaotic and dangerous,especially for pedestrians) but this does not justify theindiscriminate choice of such devices rather than other,possibly more effective, remedies.

The lack of statistically significant differences be-tween the speed values observed, in the same street, atthe speed bump or quite far from it, suggests thatprobably a combined effect of poor efficiency of thedevice and inurement of drivers to vibrations and noise,is present. Thus, the use of a number of consecutivewider devices would probably better discourage usersfrom crossing the undulation at high speeds (unlessthey wished to destroy their vehicle) and, as a long termeffect, would probably teach drivers to maintain aconstant, acceptable speed along the entire street.

Another important question to discuss is, are speedbumps really effective in protecting a crosswalk? Theimportance of this issue is related to the behaviour oflocal authorities; who tend to place a speed bump closeto any existing crosswalk; at the same time justifyingthis choice with a presumed effect of the device thatwas never really estimated from an objectivestandpoint.

On the basis of the data collected in this study, if itis true that no significant differences were found onspeed levels between free and bump sections, the sameabsence of significance was assessed in the short spatialdistance between bumps and crosswalks. Thus, it ispossible to deduce that speed bumps elicit no decisivereaction from drivers approaching a crosswalk.


Another point to underscore concerns the choice of thedistance between bumps and crosswalks: although it isreasonable to state that a distance of 20 m affords an easyview of a pedestrian starting to cross a street (or whointends to do so); if the speed over the bump is in therange of 45–72 km/h (as is the case for the majority ofthe drivers) 20 m are sufficient to allow safe braking atthe lower speed values (45 km/h); but may be inadequateat the highest levels. Thus, two opposite effects shouldbe considered and further investigated: on the one hand,there is the need to place the crosswalk inside the possiblezone of influence of the device (which means beforedrivers resume the speed they were driving at prior tocrossing the bump) but, on the other hand, a reductionin the bump-crosswalk distance will also result in areduction of stopping distance, thus making it necessaryto reduce vehicular speed to under the values observedin this study.

As regards the ‘speed-lane width’ relationship, it isinteresting to note that streets in which the lowest speedvalues were recorded are mainly characterised by theworst geometrical features and this, as expected, is a signof the stronger influence of lane width than the presenceof the speed bump on drivers’ behaviour.

Comparison between ‘bump’ and ‘no bump’ regres-sions reveals only a slight difference in the slopes of thelines, thus making it difficult, even from this point ofview, to find a noteworthy effect of speed bumps ondrivers’ speed. On the basis of world experiences, areduction of speed in the range of 20–30 km/h isdesirable after the application of the undulations while,in our case, this reduction would be theoreticallyachieved only for very large lane widths (and correspond-ingly high speed levels). Of course it is obvious that it isnot so important to obtain a decrease from 80 to 60 km/hat some sites due the presence of the bump, as it is tothan guarantee a generalised control of speed at valuesunder 40 km/h at all installations.

Finally, as shown in the speed profile analysis, the useof single devices in a certain street (rather than a seriesof consecutive well spaced bumps) shows a major weak-ness mainly in the limited spatial effectiveness of thedevice. Speed values appear almost constant both beforeand after the device location, except for a small regionwhere all the effect of the bump on drivers is concen-trated, thus giving the strongest impact on vehicle traffic.The results of such a test, together with the informationextracted from the previous analysis, reveal that theeffectiveness of speed bumps can be defined as ‘puncti-form’, since the objective of reducing vehicle speed isachieved in practice only at the point of the device. Ofcourse, further investigations are needed, possibly withan extension of the data base of speed profiles by testinga large number of installations. But in any case, such atrend was found more or less evidently at all sampledsites.

5. Conclusions

Italy adopted the use of undulations as a passive speedreduction method only 10 years ago. Ignoring manyforeign experiences (especially in UK and US) the onlytype of device installed was the so called ‘speed bump’,which is characterised by a small width compared with‘speed humps’, to date widely employed in other coun-tries. The effectiveness of such a device has been assessedby a number of studies, even though transversal sizegreatly influences the amount of speed reduction achiev-able (i.e. wider undulations are more effective).

Since no regular ‘before and after’ studies were evercarried out, this paper proposes a speed analysis of 23speed bump installations in the city of Cagliari, for themain purpose of seeing if current speed levels at thedevices and at the crosswalks protected by them arecompatible with the posted speed limits and, more ingeneral, with the safety needs of pedestrians.

Results show that in almost a third of the installations,speed limits are violated and, in another third, the 85thpercentile of speed lies in the 45–50 km/h range. Acomparison between speed values recorded at the speedbump and at the crosswalks following them show that,in most cases, the employment of such a device is notreally effective in protecting crosswalks.

On collecting speed data in sections of the selectedstreets where the effect of the speed bumps can beconsidered exhausted, results also show a general trendof drivers to maintain high speeds, while no significantdifferences were found between bump and free sections.

To arrive at a better assessment of such an effect, thespeed profile in a range of 200 m was recorded at foursites: in this case, results show that the effect of the speedbump can be considered quite local, which means thatspeed reduction is concentrated in a short distance ofabout 30–60 m.

In conclusion, an experimental relationship betweenspeed and street width was found by coupling ‘speed–width’ data both for speed bump locations and freesections. The results clearly denote an increase in speedas available lane width increases, with only slight differ-ences between the two situations. This suggests thatdrivers’ speeds are more influenced by the geometricalfeatures of the streets rather than the presence of bumps.

Although speed bumps seem to affect drivers’ be-haviour to some extent, their effectiveness as speedreducing devices is quite far from optimal and thus thechoice of local authorities to install a large number ofthem to achieve a generalised increase in pedestriansafety does not appear to be justified.

Therefore, as has already been done in othercountries, Italian authorities should plan either the useof humps instead of bumps or perhaps integrate the useof such tools with other sorts of traffic calmingtechniques.


References

Barbosa, H., Tight, M.R., May, A.D., 2000. A model of speedprofiles for traffic calmed roads. Transportation Research Part A34, 103–123.

City of Portland, Bureau of Traffic Management, 1998. City ofPortland speed bump peer review. Portland, OR, USA.

Fildes, B.N., Lee, S.J., 1993 The speed review: road environment,behaviour, speed limits, enforcement and crashes. Report CR127,Federal Office of Road Safety, Road Safety Bureau, Roads andTraffic Authority of New South Wales, Canberra, Australia.

Hodge, A.R., 1993 Speed control humps-A trial at TRL, ProjectReport 302. Transport Research Laboratory, UK.

Kassem, E., Al-Nassar, Y., 1982. Dynamic considerations of speedcontrol humps. Transportation Research Part B 16, 291–302.

Layfield, R.E. Parry D.I., 1998. Traffic calming-Speed cushionschemes, TRL Report 312. Transport Research Laboratory, UK.

Mak, K.K., 1986. A further note on undulation as a speed controldevice. Transportation Research Record 1061, 13–20.

Sumner R., Baguley C. 1979 Speed Control Humps on ResidentialRoads, TRRL Laboratory Report 878. Transport and RoadResearch Laboratory, UK.

Vis, A.A., Dijkstra, A., Slop, M., 1992. Safety effects of 30 km/hzones in the Netherlands. Accident Analysis and Prevention 24,75–86.

Watts G.R. 1973 Road humps for the control of vehicle speed, TRRLLaboratory Report 597. Transport and Road Research Labora-tory, UK.

Webster D.C. 1998 Traffic calming-public attitude studies: a literaturereview, TRL Report 311. Transport Research Laboratory, UK.

Webster D.C., Layfield R.E. 1998 Traffic Calming-sinusoidal, ‘H’ and‘S’ humps, TRL Report 377. Transport Research Laboratory,UK.

Further Reading

Broadbent, K., Salmon, A.M., 1991. An experiment with roadbumps. Highway and Transportation 11, 5–8.

Chadda, H.S., Cross, S.E., 1985. Speed (road) bumps: issues andopinions. Journal of Transportation Engineering 111, 412–418.

City of Toronto City Council., 1997, Installation of Speed Bumps onCity Streets. Policy Report, Toronto, Canada.

Clement J.P., 1982 Speed Humps and the Thousand Oaks Experi-ence, Traffic Engineering Division, City of Thousand Oaks, CA,USA.

Dahlerbruch, A., Rychlicki, M., Vaziri B. 1993 Speed humps: imple-mentation and impact on residential traffic control. Compendiumof technical papers, ITE District 6, Annual Meeting, pp. 265–270.

Engel, U., Thomsen, L.K., 1992. Safety effects of speed reducingmeasures in Danish residential areas. Accident Analysis and Pre-vention 24, 17–28.

Halbert, G., Marabian, L., Yousef, H., Murray T. 1993 Implementa-tion of a residential traffic control program in the city of SanDiego. Compendium of Technical Papers, ITE District 6 AnnualMeeting, pp. 53–60.

ITE (Institute of Transportation Engineers) Council Speed HumpsTask Force TENC-5TF-01, 1997 Guidelines for the design andapplication of speed humps. ITE, USA.

Smith, D.E., Giese, K.L., 1997. A Study on Speed Humps. Centre forTransportation Research and Education, Iowa State University.

Stephens, B.W., 1986. Road humps for the control of vehicularspeeds and traffic flow. Public Roads 50, 82–90.

Zaidel, D., Hakkert, A.S., Pistiner, A.H., 1992. The use of roadhumps for moderating speeds on urban streets. Accident Analysisand Prevention 24, 42–56.

.

do speed bumps really decrease traffic speed? an italian experience

Documents