Assessment of the Road Safety Performance of Four Roundabouts in Kumasi, Ghana

International Journal of Injury Control and Safety Promotion

ISSN: 1745-7300 (Print) 1745-7319 (Online) Journal homepage: https://www.tandfonline.com/loi/nics20

Assessment of the road safety performance of four roundabouts in Kumasi, Ghana

Shaibu Bawa & James Damsere-Derry

To cite this article: Shaibu Bawa & James Damsere-Derry (2018) Assessment of the road safety performance of four roundabouts in Kumasi, Ghana, International Journal of Injury Control and Safety Promotion, 25:3, 239-246, DOI: 10.1080/17457300.2017.1345949 To link to this article: https://doi.org/10.1080/17457300.2017.1345949

Published online: 11 Jul 2017.

Submit your article to this journal

Article views: 164

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=nics20 INTERNATIONAL JOURNAL OF INJURY CONTROL AND SAFETY PROMOTION, 2018 VOL. 25, NO. 3, 239–246 https://doi.org/10.1080/17457300.2017.1345949

Assessment of the road safety performance of four roundabouts in Kumasi, Ghana

Shaibu Bawa and James Damsere-Derry Trafﬁc and Transportation Division, Building and Road Research Institute, Kumasi, Ghana

ABSTRACT ARTICLE HISTORY Upgrade of the Ejisu-Kumasi section of the N6 saw the construction of roundabouts at selected Received 10 October 2016 intersections. Their use appears challenging particularly for vehicles with elevated Centre-of-Gravity. The Accepted 18 June 2017 ﬁ objective of this study was to establish the relative risk of traf c accident deaths or hospitalized injuries at KEYWORDS the roundabouts. Five-year crash data covering a 100 m stretch upstream and downstream of the Modern roundabout; road roundabouts were analysed. A total of 119 crashes occurred with 18% casualty hospitalization and 12% safety; road injury casualties; fatalities. Compared with the Ejisu roundabout, fatal or hospitalized injuries were signiﬁcantly higher at the rollover collisions; heavy Boadi (OR = 25.6, p = 0.0021) and Kwame Nkrumah University of Science and Technology Police Station goods vehicles Junctions (OR = 48.2, p < 0.001). Occupants of buses, heavy goods vehicles and riders of two wheelers were at elevated risk of deaths or serious injuries. Crash severity at the round abouts was aggravated by night-time, steep gradients and adverse cambers. It is recommended that, signalization and grade separation should be used in the current arrangement where appropriate.

Introduction 2014). Nevertheless, improperly designed roundabouts have adversely affected road safety especially vulnerable road users Speeding is an important risk factor for road traffic accidents such as moped riders, motorcyclists, pedestrians and older and the attendant injury severities. Nevertheless, speeding is drivers (Daniels, Brijs, Nuyts, & Wets, 2010; De Brabander & pervasive in Ghana to the extent that over 90% of motorists in Vereeck, 2007; Lord, van Schalkwyk, Chrysler, & Staplin, settlements in Ghana typically exceed the posted speed limit of 2007). Geometric characteristics which permit speeding on the 50 km/h (Damsere-Derry, Afukaar, Donkor, & Mock, 2008; approach and entry as well as sudden changes in features such Derry, Afukaar, Donkor, & Mock, 2007). The grotesque perva- sion of speeding in Ghana is largely attributable to the laxity of as improperly engineered crossfall and radius are some of the enforcement of traffic laws in general and speeding in particu- primary causes of road accidents particularly rollovers at lar. This scenario has resulted in many speeding-related road roundabouts (cited in Hall, 2014). Additionally, the lateral crashes in Ghana. In fact, over 60% of all pedestrian fatalities weight transfer in adverse cambered highways curves may in Ghana are attributable to speeding (Damsere-Derry, Ebel, aggravate accidents at roundabouts (Granlund, Haakanes, & Mock, Afukaar, & Donkor, 2010). Speed calming measures are Ibrahim, 2014). According to Taekratok (1998), roundabouts therefore imperative on the trunk roads in Ghana to improve may not be suitable in given circumstances. First, where a satis- fi upon road safety at potentially high traffic conflict zones such factory geometric design cannot be provided due to insuf cient as at junctions. space or unfavourable topography or unacceptable cost of con- One of such traffic calming measures which has been widely struction such as service relocation and property compensa- fi fl used elsewhere in the world to mitigate against traffic conflicts tions. Second, where traf c ows are unbalanced with high at priority intersections and improve upon the road safety at volumes on one or more of the approaches and some vehicles such intersections is modern roundabouts. One of the main may experience long delays, roundabouts may be a bad alterna- advantages of roundabout construction includes the reduced tive. Third, where the modal mix of vehicles on the road entails number of conflict points compared with uncontrolled inter- frequent usage of large or over-dimensional vehicles, round- sections (Mauro, Cattani, & Guerrieri, 2015; Taekratok, 1998; abouts may not be helpful (see Taekratok, 1998). In other to Tollazzi, Mauro, Guerrieri, & Rencelj,¸ 2016). Also, modern improve upon their performance, roundabouts should be visi- roundabouts in particular engender slower speeds because of ble to motorists. Since the geometric design of a roundabout is their constraining geometric characteristics which tend to essentially different from other types of traffic controls, it is reduce accident severities (Taekratok, 1998). Properly designed important for motorists to approach roundabouts with care. roundabouts are associated with significant reductions in road Apparently, some conditions for locating roundabouts have accident injury severity (Jensen, 2013; Persaud, Retting, been violated with regards to the modern roundabouts in Garder, & Lord, 2001; Qin, Khan, Bill, & Noyce, 2011). Due to Kumasi. In particular, varying unfavourable topography cou- their safety and capacity benefits, roundabouts have been pled with fairly large proportions of vehicles with elevated cen- widely used in many countries such as the USA (Rodegerdts, tre-of-gravity (CoG) (e.g. buses, heavy and long goods

CONTACT James Damsere-Derry [email protected] © 2017 Informa UK Limited, trading as Taylor & Francis Group 240 S. BAWA AND J. DAMSERE-DERRY vehicles) have in one way or the other aggravated the safety sit- Further description uation of these junctions. Therefore, we hypothesize that there Since the roundabouts are as-built, the geometric characteris- will be significant differences in the proportion of fatal and tics are slightly different in dimension and are further hospitalized casualties at each roundabout. The main objective explained below. of this research is to establish the relative risk of death or serious road injuries at the four roundabouts in Kumasi. The KNUST Police Station Roundabout The KNUST Police Station Roundabout is the first junction to Methodology be encountered from the Kumasi city (KNUST Junction) direc- tion. Characteristically, the Police Station Roundabout is a per- The study covered the section of the N6 trunk road between fect crossroad where a minor single carriageway intersects a the Kwame Nkrumah University of Science and Technology major dual carriageway at-grade. (KNUST) junction and Ejisu, a satellite town near Kumasi. The From the Kumasi approach, the 7.0 m dual carriageway road serves sub-urban conurbation communities of mixed land tapers into a single carriageway of 3.5 m at the neck of the use developments mainly of residential and commercial in roundabout and flares out on a lane width of 4.5 m. The geo- fi character. The traf c comprises of local, intra-city, long dis- metric characteristic on the Kentinkrono (Accra) bound section tance transport within the country and international transport is the exact replica of the KNUST Junction (Kumasi city) bound either operating within the city of Kumasi, or traveling up as described earlier. The entry and exit lane widths on both northern Ghana or to neighbouring Burkina Faso, Niger and minor road are 3.5 and 4.0 m, respectively (see Figure 1(a) for fl Mali. Vehicular eets comprise passenger cars, medium buses, further geometric characteristics). As illustrated in Figure 1(b), fi fi large buses and cargo trucks. In 2014 traf c volume classi ca- two gradients are discernible at the KNUST Police Station fi tion, medium to heavy vehicles constituted 12% of all traf c Roundabout. From 100 m away on the Kentinkrono (Accra) which traversed this section of the road (Ghana Highway approach to the roundabout, the road slopes at 1.5% gradient fi Authority, 2014). Uniform traf c volume is assumed within and 4.1% from the roundabout to 100 m away on the Kumasi the study site because within the 12 km stretch, the Ghana city bound. The abrupt kink at the junction presents a challenge fi Highway Authority has only one traf c count station at the for heavy goods vehicles which spontaneously breakdown at the Oduom railway crossing. This road section was recently neck of the roundabout. upgraded from a two-way single carriageway to multilane carriageway to cater for rising traffic volumes. As a result, the performance of some four priority junctions which were The Boadi Junction Roundabout converted to roundabouts has come under public criticism for The Boadi Junction Roundabout is a three-legged junction poor safety record. where a minor road (a single carriageway) intersects a dual carriageway at-grade. The dual carriageway tapers off from about 80 m on the KNUST Junction (Kumasi city) approach to a single lane of 3.5 m upon entry into the roundabout while the exit General characteristics of the study sites to the Oduom (Accra) bound flares out with a 4.1 m road The four sites targeted for this research were all retro-fitted width. Similarly, the 7.0 m wide dual carriage narrows to a roundabouts, which were converted from stop controlled inter- 3.5 m on the Accra approach and flares out with a lane width sections to modern roundabouts. All the sites have one lane in of 4.5 m on the Kumasi bound. The entry and exit lane widths the circulatory roadway of 6.8 and 2.8 m cobblestoned inner on the minor road are 4.5 and 3.5 m, respectively (see Figure 2 aprons meant to support long vehicles’ manoeuvrability and (a) for details). This roundabout is located on a slope. yield signs on entry. Lane narrowing was used on the As shown in Figure 2(b), the average gradient of the dual approaches to slow drivers through the roundabouts and flared carriageway at the Boadi Junction Roundabout within 100 m lanes at exits to speed up traffic flow. The Ejisu and KNUST radius of the roundabout is 3.3%. roundabouts have four approach legs each (including two multilane approach legs) whilst Oduom and Boadi Junction The Oduom Roundabout Roundabout have three approach legs each (including two multilane approach legs). Roundabouts, particularly modern Just like the Boadi Junction Roundabout, the Oduom Round- roundabouts are quite uncommon in Ghana. Therefore, the about is a three-legged junction where a single carriageway four roundabouts under investigation were only recently, i.e. intersects a dual carriageway at-grade. The Oduom Round- 2005 constructed. Characteristically, the modern roundabouts about is an exact replica of the Boadi Junction Roundabout. are uniquely different from the few existing ones in Ghana Dimensions of the geometric features at the Oduom Round- because they have truck apron around the central circles. The about such as entry and exit lane widths on the major and apron is a textured concrete surface next to the outside curb of minor roads are the same as those at the Boadi Junction the central island designed to allow large trucks to turn with (see Figure 3(a)). their rear wheels leaving the roadway and riding up onto the The roundabout is situated on a crest. Therefore, two local apron area. gradients were discernible. The local gradient for 100 m on the INTERNATIONAL JOURNAL OF INJURY CONTROL AND SAFETY PROMOTION 241

To C.C.C

KNUST POLICE STATION

4.0 3.5 To TECH Taper 3.0 JUNCTION 68.0 1.5 Taper 6.8 7.0 7.0 4.5 R5.0 3.5 4.0 4.0 3.5 4.5 7.0 2.8 7.0 6.8 91.0 Taper 1.5 To Taper 3.5 4.0 1.5 KENTINKRONO

All dimensions are in meters Edge of Carriageway To CAMPUS / Edge of Walkway QUARTERS

(a)

-4.08%

-1.51% Roundabout

Existing Level 99.64 99.15 99.17 98.88 98.80 98.59 98.41 98.29 98.15 98.03 97.96 101.04 104.37 103.95 103.63 103.18 102.80 102.43 102.07 101.64 101.28 100.29 100.27 100.00 99.53

(b)

Figure 1. (a) Plan and (b) longitudinal section of the KNUST Police Station Roundabout.

Fumesua (Accra) bound approach is 1.6% while the gradient There is a layby on the Kumasi approach which intersects from the roundabout to 100 m on the KNUST Junction the minor road leading to the Ejisu township. It is important to (Kumasi city) bound is 1.3% (see Figure 3(b) for details). indicate that while the minor road from the Ejisu township intersects the roundabout at a tangent, the road to Efiduasi/Eji- suman Senior High School intersects the roundabout perpen- dicularly (see Figure 4(a) for further details). Two slopes were The Ejisu Roundabout also deduced from the vicinity of the roundabout. While 100 m away on the Accra bound has a slope of 0.8%, the Kumasi Just like the KNUST Police Station Roundabout, the Ejisu bound has a gradient of 1.6% (Figure 4(b)). Roundabout is a crossroad where a single carriageway intersects a major carriageway at-grade. From the roundabout, the major road is all dual carriageway on the Kumasi city bound Study period and data sources and a single carriageway on Accra bound. The entry lane width The study period spans five years, i.e. from 2009 to 2013, after from the Accra approach is 3.6 m while the Kumasi approach the road was upgraded and opened to traffic. At each location, is 3.7 m wide and the exit lanes on the Accra and Kumasi casualty data within 100 m upstream and downstream of the approaches are 4.7 and 4.8 m wide, respectively. Like the roundabouts were used. Accident data was obtained from Oduom Roundabout, the Ejisu Roundabout is located on a the accident data bank of the Building and Road Research crest. Institute (BRRI), Ghana. The data is stored in a computer 242 S. BAWA AND J. DAMSERE-DERRY

Taper To TECH Taper 69.2 JUNCTION 1.5 84.4 6.8 7.0 7.0 4.5 R5.0 3.5 4.0 4.0 3.5 4.1 7.0 2.8 7.0 6.8 80.0 1.5 To ODUOM Taper 3.5 4.5 1.5 66.5 Taper

SHELL FILLING All dimensions are in meters STATION Edge of Carriageway Edge of Walkway To BOADI

(a)

3.34%

Roundabout

Existing Level 95.61 95.99 96.20 96.58 96.80 97.16 97.59 97.90 98.21 98.60 98.89 99.41 100.18 100.30 100.34 100.88 101.25 101.58 101.97 102.37 102.71 103.00 103.39 103.77 103.95

(b)

Figure 2. (a) Plan and (b) longitudinal section of the Boadi Junction Roundabout.

software called Micro-Accident Analysis Package developed by Classification of casualty injury severity in Ghana the Transport Research Laboratory , UK. The data is geo-coded Casualty injury severity in Ghana is classified into four groups. to correspond with the artificial and natural landmarks of each The classification of the casualty severity is guided by a 30-day road section. The source of the accident statistics is from the timeline after the crash. The categories are fatal, serious inju- police who collect primary information from accident scenes, ries, minor injuries and no injuries. A fatal casualty injury for accident victims and eyewitnesses. Some underreporting is instance occurs when an accident victim dies within 30 days inherent in the accident statistics in Ghana. According to Salifu after the injury. Similarly, a serious casualty injury is a situation and Ackaah (2009), non-reporting of road accidents in Ghana in which a road user who is injured in a crash is hospitalized ranges between 57% of property damage collisions to 0% of for at least 24 h within 30 days after the incident. Minor inju- fatal crashes. This report further indicates that there is a preva- ries are characterized by slight injury in which the accident vic- lence of 27% of under-recording in the BRRI accident statistics. tim was treated and discharged or not hospitalized for at least N.B: Under-recording is the difference between the reported 24 h within the 30-day period following the accident. crashes in the police stations and the actual data obtained by the BRRI field data collectors. Observation Vehicles’ manoeuvrability was observed at the roundabouts to ascertain their functionality and capacity.

Study design Analysis The study design used in this research is the case-series report. Descriptive statistics were initially used to present the data in the The construction of modern roundabouts is a fairly new intro- form of conditional probabilities involving the dependent vari- duction in Ghana and to the best knowledge of the authors, able (i.e. casualty injury severity) and the independent variables, there are only four of their kind. However, there is an urgent namely; vehicle type, time of day, location (study site) and colli- need to assess the performance of these limited modern round- sion type. For the second phase of analysis, an inferential statis- abouts to guide their future replication in similar road environ- tics based on a multivariable and binary logistic regression using ments in Ghana. a dichotomous-dependent variable and dummy variables of the INTERNATIONAL JOURNAL OF INJURY CONTROL AND SAFETY PROMOTION 243

Apron 1.5 Taper To TECH Taper 69.0 JUNCTION 1.5 80.0 6.8 1.5 Walkway 7.0 7.0 4.5 R5.0 3.5 4.0 Median 4.0 Median 4.0 3.5 4.0 7.0 2.8 7.0 6.8 80.0 1.5 To FUMESUA 69.0 Taper 3.5 4.5 1.5 Taper

All dimensions are in meters Edge of Carriageway Edge of Walkway To ANWOMASO

(a)

1.31% -1.62%

Roundabout

Existing Level 99.77 98.47 98.57 98.67 98.81 98.91 99.02 99.22 99.30 99.39 99.65 99.40 99.76 99.74 99.86 99.84 99.45 99.31 99.17 99.09 98.89 98.84 98.76 98.58 98.54 98.22

(b)

Figure 3. (a) Plan and (b) longitudinal section of the Oduom Roundabout. independent variables was used. Post-estimation marginal prob- As shown in Table 1, pedestrian casualties were more likely abilities were computed and the model adequacy was checked to endure serious or fatal injuries as the vast majority (80%) of using Hosmer–Lemeshow goodness-of-ﬁt. Crude and adjusted all pedestrian casualties were either fatal or serious injuries, n = odds ratios were used as surrogates for relative risks. Data was 15. Casualties in rollover collisions also resulted in 46% fatal or analysed using Stata version 13. serious casualty injuries, n = 13. Casualties in rear-end

Results Table 1. Distribution of injury severity by casualty attributes. Casualty injury severity Over the study period, 119 casualties occurred at the study Time of day No/minor Serious/fatal Total locations out of which 12% was fatal, 18% was serious injuries Daytime (6:00–17:59) 64 (75%) 21 (25%) 85 (100%) which resulted in hospitalisation, 30% was slight injury and Night-time (18:00–05:59) 19 (56%) 15 (44%) 34 (100%) 40% resulted in no injury (property damage only). Among Location (roundabout) the roundabouts investigated, KNUST Police Station was the Ejisu 9 (75%) 3 (25%) 12 (100%) location where the highest serious or fatal casualties (39%) Boadi 21 (68%) 10 (32%) 31 (100%) KNUST Police Station 35 (71%) 14 (29%) 49 (100%) occurred followed by Boadi (28%) and Oduom (25%) round- Oduom 18 (68%) 9 (33%) 27 (100%) abouts while the least serious or fatal casualty injuries (8%) Vehicle type occurred at the Ejisu Roundabout. Generally, 30% of all casu- Cars 43 (86%) 7 (14%) 50 (100%) alties which occurred at the study locations were serious or Bus 17 (55%) 14 (45%) 31 (100%) fatal injuries. As illustrated in Table 1, the proportion of casu- Two wheelers 2 (29%) 5 (71%) 7 (100%) Heavy goods vehicles (HGV) 19 (70%) 8 (30%) 27 (100%) alties which sustained serious or fatal injuries at night-time Other 2 (50%) 2 (50%) 4 (100%) (44%) was statistically higher than the proportion which sus- Collision type tained similar injuries during the daytime (25%), p = 0.0186. Rear-end 47 (80%) 12 (20%) 59 (100%) Also, a sample test of two proportions indicated that casual- Sideswipe 12 (86%) 2 (14%) 14 (100%) Ran-off 13 (54%) 2 (2%) 15 (100%) ties who were private car occupants and sustained serious or Rollover 7 (54%) 6 (46%) 13 (100%) fatal injuries (14%) were statistically lower than the propor- Hit pedestrian 3 (20%) 12 (80%) 15 (100%) tion of bus casualties (45%), p = 0.0009 and heavy goods vehi- Others 1 (33%) 2 (67%) 3 (100%) cle occupants (30%), p = 0.0492, Table 1. Total 83 (70%) 36 (30%) 119 (100%) 244 S. BAWA AND J. DAMSERE-DERRY

To JUABEN / EFFIDUASE 1.5 1.5

Taper To KUMASI 32.0 1.5 26.5 6.8 Not Dualized 8.0 4.8 R5.0 3.6 3.5 4.0 Median 2.8 3.5 8.0 3.7 4.7

22.6 6.8 Dualized Section 16.2 To ACCRA Taper

All dimensions are in meters 2.3 Edge of Carriageway Edge of Walkway

To EJISU TOWN

(a)

-0.79% 1.59%

Roundabout

Existing Level 98.51 98.68 98.94 99.10 99.32 99.48 99.58 99.68 99.78 99.94 99.78 99.42 99.86 99.94 99.90 99.79 99.75 99.70 99.58 99.54 99.39 99.59

(b)

Figure 4. (a) Plan and (b) longitudinal section of Ejisu Roundabout. collisions which constituted 50% of all the casualties resulted in The relative risk of death or serious injuries varies across the 20% serious or fatal injuries. locations (roundabouts). Compared with the Ejisu Round- The second part of the analysis was a binary logistic regres- about, the relative risk of death or serious injuries is signifi- sion and the results are illustrated in Table 2. cantly higher at Boadi Junction Roundabout (AOR = 25.5, p = 0.021) and KNUST Police Station Roundabout (AOR = 48.2, Table 2. Predictors of accident severity at selected modern roundabouts in p = 0.008). Compared with private car occupants, the relative Kumasi. risk of death or serious injuries was significantly higher among Odds ratios bus occupants (AOR = 18.8, p = 001) and only marginally Variable Crude Adjusted p-value 95% CI higher among heavy vehicle occupants (AOR = 4.6, p = 0.087), Time of day see Table 2. The relative risk of death or serious injuries was Daytime (6:00–17:59) 1.00 1.00 –– fi Night-time (18:00–05:59) 2.41 10.56 0.001 2.74–40.77 signi cantly higher at night-time compared with daytime Location (roundabout) (COR = 2.4, AOR = 10.6, p = 0.001). Ejisu 1.00 1.00 –– The predicted probabilities at the various locations show Boadi 1.43 25.58 0.021 1.61–405.07 that if all the casualty injuries which occurred at the selected KNUST Police Station 1.20 48.18 0.008 2.75–845.27 Oduom 1.50 10.50 0.117 0.55–199.24 sites within the study period had occurred at Ejisu Round- Vehicle type about, 11% of the accident victims would have died or suffered Cars 1.00 1.00 ––serious injuries, p = 0.010. Also, treating all casualties as though Bus 5.06 18.84 0.001 3.48–102.13 they occurred at the Oduom Roundabout would have resulted – Two wheelers 15.36 52.14 0.003 3.67 740.91 in 24% death or serious injuries, p < 0.001. On the other hand, Heavy goods vehicles 2.56 4.57 0.088 0.80–26.18 Other 6.14 10.79 0.230 0.22–522.0 the marginal probability of death or serious injuries at Boadi Collision type Roundabout 32%, p < 0.001 and KNUST Police Station Rear-end 1.00 1.00 ––Roundabout 39%, were on the higher side, see Figure 5. The Sideswipe 0.65 0.40 0.419 0.04–3.73 marginal probability of death or serious injuries at the Oduom Ran-off 0.60 17.53 0.087 0.66–465.09 Rollover 7.83 1.13 0.900 0.17–7.49 Roundabout was not statistically different from the marginal Hit pedestrian 15.67 596.23 0.000 27.91–12730.60 probability of death or serious injuries at the Ejisu Roundabout Others 3.36 7.16 0.021 1.34–38.70 F(1, 119) = 2.45, p = 0.1174. In contrast, the marginal probabil- Constant 0.00060 0.000 0.00002–0.02 ity of casualty death or serious injuries at the Boadi Junction INTERNATIONAL JOURNAL OF INJURY CONTROL AND SAFETY PROMOTION 245 .5 .4 .3 .2 .1 Probability of casualty death or serious injuries serious or death casualty of Probability 0

ejisu boadi knust oduom Location of roundabout

Figure 5. Mean casualty death or serious injuries by location (roundabout).

Roundabout was significantly higher than the probability of Roundabout, p = 0.008 compared with the Ejisu Roundabout. death or serious injuries at the Ejisu Roundabout F(1, 119) = The Boadi and KNUST Police Station Roundabouts have been 5.29, p = 0.0214. Similarly, the marginal probability of death or scenes of several fatal or serious casualty injuries largely serious injuries at the KNUST Police Station Roundabout was because of their faulty designs. In particular, adverse camber significantly higher than those of Ejisu Roundabout F(1, 119) = can lead drivers to be drawn to the edge of the roads leading to 7.03, p = 0.008. This probably depicts locational anomalies at accidents (Bonham, 2014). According to the French guideline, the KNUST Police Station and Boadi Junction Roundabouts installing a roundabout on a roadway with a slope of 3% or compared with the Ejisu Roundabout. more may be problematic. This report further indicates that Post-estimation Hosmer and Lemeshow goodness-of-fit test some conditions such as reverse superelevation on the circulat- was conducted to check the model adequacy. The test statistic ing roadways or locating the roundabout on the crest of a verti- x2(8, N = 119) = 8.10, p = 0.4235 was not statistically signifi- cal curve of one of the intersecting roads may compromise on cant indicating that the model was a good fit. safety (Taekratok, 1998). These fundamental engineering con- It was observed that rampant traffic jams are associated with siderations have been flaunted at the Boadi and KNUST Police these junctions during peak hours. During peak hours, vehicle Station in particular. The precipitous gradient at the Boadi queues may extend beyond 500 m and move at level of service Junction Roundabout and the unfavourable sudden kink at the of F. During the morning peaks (07:00–09:00 GMT), the grid- KNUST Police Station Roundabout give credence to the worse locks are built on the Kumasi bound approach while during road safety performance of these junctions. The kink at the the evening peak hours (17:00–19:00 GMT), the long queues KNUST Police Station Junction also poses challenge to heavy are built on the Accra bound approach. trucks which breakdown spontaneously at the neck of the roundabouts thereby compounding the gridlock at this location. Discussion and conclusion Given the backdrop that modern roundabouts are uniquely In this evaluation, it is notable that these roundabouts have not new in the country, their installation should be heralded by performed optimally as 30% of all casualties at these locations educating motorists regarding the modus operandi of modern resulted in fatalities or serious injuries. A properly designed roundabouts through mass media education, videotapes and roundabout is characterized by a possible eradication of fatal television. This approach was used successfully in Florida, Ver- or serious injuries as reported by Qin et al. (2011). Therefore, mont and Maryland in the USA to improve upon the public in a situation whereby as high as 30% of injured casualties at perception of roundabouts and improve upon their safety these roundabouts resulted in death or serious injuries is a (Taekratok, 1998). As noted earlier, the two lane approaches to safety concern. The relative risk of death or serious injuries at the roundabouts narrow to single lanes upon entry and the Ejisu Roundabout was not statistically different from the throughout the roundabouts. Without traversing the cobble Oduom Roundabout, p = 0.117. However, the relative risk of stones, heavy goods vehicles and large buses which have ele- death or serious injuries was significantly higher for the Boadi vated CoGs manoeuvre these tight corners of roundabouts Roundabout, p = 0.021 and the KNUST Police Station with extreme difficulties leading to frequent accidents 246 S. BAWA AND J. DAMSERE-DERRY particularly, lateral displacements of heavy vehicles. Given the Disclosure statement novelty of modern roundabouts in Ghana, motorists should No potential conflict of interest was reported by the authors. have been adequately educated on the use of these facilities. The fact that motorists were not adequately informed about their usage has exacerbated the road safety situation especially References frequent rollovers of HGVs and large buses astride the Bonham, S. (2014). Woman injured after skidding off roundabout. Cay- roundabout. man Compass. Existing physical conditions such as approach steep slopes Damsere-Derry, J., Afukaar, F.K., Donkor, P., & Mock, C. (2008). Assess- and inadequate right of way at some of the sites may have ment of vehicle speeds on different categories of roadways in Ghana. exerted limitations on the size of the roundabouts in general. International Journal of Injury Control and Safety Promotion, 15(2), 83–91. The KNUST Police Station and Boadi Junction Roundabouts Damsere-Derry, J., Ebel, B., Mock, C., Afukaar, F., & Donkor, P. (2010). for instance were constrained in terms of approach grades and Pedestrians injury patterns in Ghana. Accident Analysis & Prevention, insufficient banking which adversely affect manoeuvrability of 42, 1080–1088. buses, long vehicles and cargo trucks in particular. Daniels, S., Brijs, T., Nuyts, E., & Wets, G. (2010). Explaining variation in Also, lane narrowing meant to guide traffic through the cir- safety performance of roundabouts. Accident Analysis & Prevention, 42(2), 393–402. culatory often leads to gridlocks because of slow pace of move- De Brabander, B., & Vereeck, L. (2007). Safety effects of roundabouts in ment leading to long traffic queues at all the sites, usually in Flanders: Signal type, speed limits and vulnerable road users. Accident the morning and evening peak hours. Some experts, e.g. Taek- Analysis & Prevention, 39(3), 591–599. ratok (1998) argue that, roundabouts are not the best alterna- Derry, J.D., Afukaar, F.K., Donkor, P., & Mock, C. (2007). Study of vehicle tive in a situation where a major road intersects a minor road speeds on a major highway in Ghana: Implication for monitoring and control. Traffic Injury Prevention, 8(2), 142–146. thereby resulting in unacceptable delays on the major roads Ghana Highway Authority. (2014). Traffic survey on Accra-Kumasi road. (Taekratok, 1998). He further argues that, roundabouts are Census point: Railway crossing, Oduom. In Authority Highway Ghana deemed inappropriate when the number of lanes on the circu- (Ed.). latory do not meet capacity requirements or if the use of the Granlund, J., Haakanes, I., & Ibrahim, R. (2014). Lowered crash risk with roundabout produces greater vehicle delay or increased diffi- banked curves designed for heavy trucks. Paper presented at the 13th International Symposium on Heavy Vehicle Transport Technology, culty for pedestrians. Currently, the roundabouts are scenes of San Luis. continuous gridlocks. The fact that the relative risk of death or Hall, T.M. (2014). Heavy vehicle rollover propensity at roundabouts on serious injuries is high at night-time suggests that improving highspeed roads. http://docs.lib.purdue.edu/open_access_theses/329 visibility of the roundabouts at night may be a road safety Jensen, S. (2013). Safety effects of converting intersections to roundabouts. benefit. Transportation Research Record: Journal of the Transportation Research Board, 2389,22–29. Lord, D., van Schalkwyk, I., Chrysler, S., & Staplin, L. (2007). A strategy to Conclusion reduce older driver injuries at intersections using more accommodat- ing roundabout design practices. Accident Analysis & Prevention, 39 Generally, roundabouts are among the safest locations on the (3), 427–432. roadways. Nevertheless, the four modern roundabouts evaluated Mauro, R., Cattani, M., & Guerrieri, M. (2015). Evaluation of the safety performance of turbo roundabouts by means of a potential accident rate intheKumasiMetropolishavefailedtoalargeextent.Thefol- model. The Baltic Journal of Road and Bridge Engineering, 10(1), 28–38. lowing engineering facilities are recommended to improve upon Persaud, B., Retting, R., Garder, P., & Lord, D. (2001). Safety effect of the road safety at these junctions. Instead of the roundabouts, roundabout conversions in the United States: Empirical Bayes observa- signalization at the Ejisu and Oduom Roundabouts will be tional before-after study. Transportation Research Record: Journal of – appropriate while grade separation will be convenient at the the Transportation Research Board, 1751, 1 8. Qin, X., Khan, G., Bill, A., & Noyce, D.A. (2011). Comprehensive safety Boadi and KNUST Police Station Roundabouts. Finally, with evaluation of roundabouts in Wisconsin. Journal of Transportation the adverse road safety performance associated with the installa- Safety & Security, 3(4), 289–303. tion of modern roundabouts in Kumasi on the trunk roads, it is Rodegerdts, L. (2014). Status of roundabouts in North America. Paper pre- recommended that their installations be minimized and if possi- sented at the 4th International Conference on Roundabouts, bleavoidedonthetrunkroadswhichhavelargeproportionsof Seattle. Salifu, M., & Ackaah, W. (2009). Under-reporting of road traffic crash data buses, trucks and goods vehicles with high CoG. in Ghana. Paper presented at the 4th IRAD Conference, Seoul. Taekratok, T. (1998). Modern roundabouts for Oregon. Salem, OR: Oregon Department of Transportation. Acknowledgments Tollazzi, T., Mauro, R., Guerrieri, M., & Rencelj,¸ M. (2016). Comparative analysis of four new alternative types of roundabouts: “Turbo”, Authors are grateful for the in-kind support received from the Building “fl ” “ ” “ fl ” and Road Research Institute (BRRI), Ghana. In particular, authors are ower , target and four- yover roundabout. Periodica Polytechn- very grateful to Mr Joseph Larbi of BRRI for his assistance in data retrieval ica Civil Engineering, 60(1), 51. from the accident database.