Comments on
Accidents at urban mini-roundabouts
TRL Report 281 by J V Kennedy (TRL) and R D Hall (University of Southampton)

Abstract  (from the report) The report gives the findings of a study of accident risk based on a national stratified sample of 200 3-arm and 100 4-arm urban mini-roundabouts on 30mph single carriageway roads (2100 personal injury accidents). Tabulations are given showing frequencies, severities and rates by type of central island and by region. The accidents are also tabulated by accident group, road user involvement and number of casualties per accident. The main objective of the study was to develop relationships between accident frequency and traffic flow, road features, layout, geometry, land use and other variables. The technique of generalised linear modelling was used to develop such relationships for different types of accidents. My comments:  Quite a proportion of the accidents at mini-roundabouts involve one or more drivers/riders who have failed to appreciate the presence of the mini-roundabout in time to respond correctly.  While this problem should come out "in the wash" statistically, I am sceptical of this approach just yet.  When drivers fail in this way, the consequences could vary considerably from nil to a fatality and the type of accident could also vary enormously.
This is a serious problem at mini-roundabouts and the statistical charts derived from the types of accidents at the sites studied seem to vary considerably from the accidents at the sites which I designed and installed mainly in Berkshire (UK). While crossing accidents featured strongly in the TRL study, these were just about the only accidents that I observed at my sites. And they featured strongly at one site in particular where I knew I had problems getting the required deflection.

The TRL sites showed accidents of many kinds which I did not seem to get; in particular:

  • Merging - I had none

  • Shunt accidents on approach - I had very few

  • Single vehicle accidents - I had very few

  • Pedestrian accidents - I had very few

These accidents are often associated with failure of drivers to appreciate the presence of the mini-roundabout which points strongly to the comments I make on the design of the approaches on my detailed design page.

The TRL report indicates that accident types at 3-arm mini-roundabouts were as follows:

Note the significant numbers of crossing accidents, but over the 200 sites as a whole the variety of accidents is surprising.

The single vehicle accidents often involve buses stopping abruptly with consequent injury to passenger(s).

Merging accidents tend to occur where drivers following a relatively straight kerbline fail to realise the presence of the junction. This approach is best split into two narrow lanes to control approaching traffic in sufficient time. If you cannot do this consider laying a buff coloured anti-skid surface, but try the two-narrow-lane idea if you can; it really does work.

Similarly shunt accidents often involve one driver reacting severely having noticed the mini-roundabout very late on approach.

[The zero thickness wedge represents crossing movements which appear very significantly in the next chart below. Both charts were derived from the same table.]

 Below is the equivalent plot of accident types at 4-arm mini-roundabouts:

Note the high proportions of crossing accidents mostly right angle crossing but also other right turn accidents.  

Otherwise the accident pattern is basically the same as for 3-arm mini-roundabouts - the overall proportions of the accident types being less as the accident types involving right angle and other crossing predominate; (the right angle crossing type not being represented at all in the three-arm scenario).

The central island at crossroads must be large enough to deflect all crossing traffic especially where there is potentially a straight path across the junction.  To achieve this it will often be necessary for the central island to be larger than the 4m currently prescribed in the UK TSRGD (signs regulations). Some authorities will not install 4-arm mini-roundabouts because of their high accident potential. This is a pity. make sure that the approaches make the junction presence clear and then ensure good deflection.

I have asked DfT to consider removing the 4m central island diameter constraint in the next TSRGD revision.
Still no go (Nov 2004)
See my page on
mini-roundabouts at crossroads
The MIDI-roundabout.

NOTE: These are RIGHT HAND DRIVE diagrams but apply just the same to UK layouts...
The conflicts illustrated here represent the various risk factors in vehicle to vehicle collisions.
The right-angle or broadside ones matter the most and these are represented by the larger red circles.

Site geometry  Strictly, it proved impossible to determine whether the junctions studied were T-junctions, Y-junctions or symmetrical (3-arm), or X-roads or K-junctions (4-arm), so TRL decided to consider movements between respective arms as the sole arbiter of direction,  i.e. Movement from arms 1-3 and 2-4 represent "right angle crossing" regardless of the actual geometry, and at 3-arm junctions - left turning means taking the next exit after entry and right turning means leaving at the second exit.
[Applies to left hand drive -  for right hand drive countries left and right should be interchanged.]

Other observations from the report:

What (from report) Why (my interpretation)
Average accident frequency at 4-arm sites was 1.35 pia per year about 50% more than average accident frequency at 3-arm sites of 0.92 pia per year. Possibly more flow at four arm minis and much more scope for accidents because there is more scope for weak design. Remember four "ahead" crossing movements instead of one or at most two at 3-arm junctions.

This is an almost meaningless statistic.Where mini-roundabouts are installed on busy road junctions the risk will normally be higher; but "busyness" may help. The busy Binfield Crossroads has a good safety record.

Average severity was 11.6% at 3-arm sites and 14.1% at 4-arm sites. Proportionately more crossing accidents which tend to be the serious ones, esp. when vulnerable users (riders of two wheeled machines) are involved. At most 4-arm sites the deflection is hopelessly inadequate owing to the 4m max. UK island size.
Mean accident rate was 12.5 PIA per 100 million vehicles inflow at 3-arm sites and 22.8 PIA per 100 million vehicles inflow at 4-arm sites. As above - more risk due to poor design, and the 4m diameter constraint.
Summary of observations 3-arm
Average accident frequency Acc/yr 0.92 1.35
Average severity 11.6 14.1
Mean accident rate PIA/100M vehs 12.5 22.8

Approach lanes: Just 6 of the 618 approaches at 3-arm sites had 2 lanes on the approach, 232 arms had more than one lane on the approach leaving 386 approach arms with only one lane marked.  At 4-arm sites there were 420 approaches of which just 87 had more than one lane marked leaving 333 approaches in single lane.

My comment on this:  I have repeatedly found that an increase in the number of approach lanes on approaching a (mini-) roundabout can be a crucial safety factor as well as a capacity benefit. Such changes are recommended in current DfT advice but this then goes on to prohibit lane widths less than 3m at the give-way line. The result is that many approaches are marked in one wide single lane which could so easily have been marked in two. At a roundabout a large vehicle occupying more than one lane does not "baulk" traffic as it almost certainly would at signals, so narrower lanes are useful; they are good for cyclists too as cyclists can occupy such a lane without drivers trying to push past.

Involvement of 2-wheelers (powered and cycles)

At 3-arm sites 39.9% of accidents involved 2-wheelers; the majority of these accidents were of the entering/circulating type. At 4-arm junctions the proportions were even higher.

2 wheeler accidents Cycles Motor-cycles
3-arm sites 22.8 17.1
4-arm sites 19.5 16.9
% entering/circulating (3-arm sites) 70.7 53.7
% entering/circulating (4-arm sites) 80.7 69.7
Relative vulnerability compared with cars
3-arm sites
4-arm sites



The vulnerability of 2-wheelers at mini-roundabouts is well known, but there remains much concern about the proportions of drivers who enter a mini-roundabout either failing to appreciate its presence or expecting little to yield to or being able to "straight-line" owing to lack of deflection; and this makes 2-wheel riders particularly vulnerable. Pedal cycles and motor cycles were 7 & 8 times more likely to be involved.

Links to other pages:

Penntraff - Dec 2014
Pages designed by: