Large or 

Roundabout under autoroute, Calais

Part of a large roundabout
under an auto-route at Calais
(measuring the outward crossfall)

Problems with large roundabouts (UK):

  • Circulation speeds much too high
  • Difficult to enter - due to high speed circulation
  • Rollover crashes of large vehicles
  • Take up too much space
  • Difficult to maintain drainage systems
  • Most large roundabouts at key nodes now signalised

We should not be building large roundabouts in the UK any more - they perform too badly...most existing ones are being signalised...

It is not the intention on this page to go into depth concerning the capacity and design details of large or normal roundabouts. This work has been done extensively elsewhere. What I am trying to do here is to point out problems associated with large roundabouts (rotaries) and how these problems may be avoided. The French have broken away from UK design rules and achieved considerable success with neat and tidy layouts that function well. We should all learn from what they have done. The Americans have compromised between the two styles. They have gone so far as possible for single lane roundabouts even having to add truck aprons to allow the largest vehicles to circulate. The truck apron is a very useful tool as it forces light vehicles to circumnavigate at larger radius so restraining circulation speed; but the truck apron itself may be responsible for the overturning of trucks at some American sites, a problem mostly not happening in France. Also the Americans sometimes use crowned circulatory roadways which I have always considered to be hazardous. However, all the roundabouts at Carmel (installed since 1999) have outward drainage and a superb accident record.


In my seminars we have looked at the operation of a number of large roundabouts and the problems associated with them. I have remarked already that large roundabouts perform poorly with traffic circulating at excessive speed resulting in loss of capacity and safety problems. There comes a point when large roundabouts are simply inefficient at best and downright dangerous at worst. The giant roundabouts associated with our motorway and trunk road interchanges leave so much to be desired and many of them have failed to handle the traffic demand placed upon them without extensive installation of traffic signals to enhance safety and/or capacity. We (TRL) knew of this as long ago as 1971 - no more excuses.

The French are not falling for our mistakes on this;
our worst problem is designing the roundabout allowing excessive speed on the circulatory roadway:

  • Because of lack of entry deflection; although guidelines introduced in 1984 have improved this

  • Because the whole roundabout is too large;

  • Because inward drainage crossfalls encourage speed.

An accident study in France indicates that outward sloping circulation is much safer.
The table below illustrates the risk of various accident types at sites which slope either way.

Relative safety of inward/outward circulating carriageways (France)

Circulating carriageway slope

Inward sloping Outward sloping
Accident type 42 roundabouts 21 roundabouts
Total accidents per roundabout year



Accidents due to loss of control at entry



Accidents due to loss of control on circulation



Accidents due to failure to give-way/yield



Note the total absence apparently of loss of control in circulation - in the UK we are very worried about this regarding HGVs (trucks) and in icy weather. There is no information on nose-to-tail accidents (shunts) on entry - a serious problem at some large UK roundabouts that I have studied.
Source: Modern Roundabout practice in the United States - A Synthesis of Highway Practice - Transport Research Board National Research Council - 1998.
I think we need to learn more from the French here and I regularly visit some of their sites in northern France. Here all new roundabouts are drained away from the central island. This has a significant effect upon circulating speeds and according to a safety study reduces accidents too. But they are making one mistake that is causing some problems and that is their adherence to single lane entry. I have been advocating two lane entries at mini-roundabouts for some years for safety reasons let alone to handle traffic demand. My trip during a busy period in August 1999 (the solar eclipse) highlighted this problem. Several roundabouts in key areas on the network had very long queues (over 1km) because of single lane entry. I have studied some of the A40 grade separated roundabouts in West London. A particular problem there seemed to be large numbers of nose to tail shunt accidents. As many as 50 accidents in a three year period at each site involved mainly shunts with a few entry/circulating accidents. The approach configurations seemed to be designed to encourage high speed right up to the give-way line with super-elevated curves on the approaches which were often single wide (4m) lanes before the approach flare. We seem to have missed out badly on medium sized roundabouts, where there is plenty of room to achieve high capacity but where the layout still imposes sufficient deflection to prevent excessive speeds in circulation. Probably one of the best in this category is the A26/A275 roundabout in Lewes, East Sussex, a three arm roundabout where the circulatory roadway has been drained outwards ensuring that the central island remains high and visible. It also includes a nice piece of environmental art. See below:


Outward draining normal and large roundabouts

General view of french roundabout
lorry negotiating french roundabout
Here are views of two roundabouts near Boulogne in northern France.  The outward draining carriageway keeps speeds noticeably slower than UK roundabouts of the equivalent diameter.  The slopes that I have measured appear to be about 1:40 but although the roundabouts are on a slope themselves the radial outfall appears to be consistent;  this means that there are no, or very slow, changes to the lateral G-forces as a vehicle circulates the roundabout. But the roundabout MUST be circular - no ellipses or other shapes. The entry to these roundabouts needs to be radial; it is here that the sharpest turn, smallest radius must occur.
This roundabout lies on a gentle slope and has not been profiled other than to follow that slope. It is at a rural location and the lighting unit on the centre is the only street light in the area. This is a classic "pillar of cloud by day and pillar of fire by night" example of the use of street lights on the central island. Excellent! Several clock-towers seem to have ended up on roundabouts; this one by Torquay harbour operates very well. Large pedestrian flows are accommodated at a split Pelican crossing through the splitter island just off the right of the picture.

Here is a graph that indicates the relative G-forces at a 60m radius roundabout with various circulatory speeds and crossfalls:

At 60m radius this is quite a large roundabout. It is generally accepted that the limit of comfort is 0.2g so around the 23-27mph area. Although the adverse crossfall as used in France (above) looks alarming, it is clear that this is a steady state condition and dynamically nothing awful is happening.

Rollover - What is going on?

Unfortunately we were all misled by a spectacular demonstration of truck rollover at TRL open days in 1971. A rigid truck was fitted with outriggers and was driven round a small diameter roundabout of about 10m radius at 13mph and then at 16mph.

At the lower speed the truck was well down on its outer suspension but stable. At 16mph the inner wheels began to lift and the truck would have rolled but for outriggers that prevented it. So what was going on?

The lateral g- force keeping a truck on a circular path is determined by the formula:

F = V2/R

where V is the speed in metres/sec and R the radius of the truck path. In this case F = 4.25 m/s2 or 0.43G at 15mph. This is already over twice the comfort limit understood to be about 0.2G. I believe this demonstrated what can happen but only if a driver is prepared to push way beyond acceptable limits. 

In short the test would not be possible without the driver being strapped in a full harness seat belt system and so is NOT representative of typical truck rollover.

This second graph of a much smaller roundabout indicates the relatively little effect of the crossfall. At 0.2g the speeds are in the range 14-17mph. Looking at the cross section of a truck it takes a lot more than 0.2g for the effective weight to fall outside the wheel-track. All the research that I have done recently (Oct 2010 onwards) indicates that several experts have looked into truck rollover and have shown that centrifugal forces cause rollover at certain speeds  depending upon the height of the Centre of Gravity (CoG) in relation to the axle width. While this is a useful guide to the relative risk of rollover I remain to be convinced that this is the correct issue. But the reason I reject centrifugal force as the principal cause needs some explanation.

In all cases of rollover crashes reported that I have found, there is a common thread - 

Drivers were NOT aware that the rollover was about to occur.

This suggests strongly that centrifugal force is not the issue - why? Because centrifugal force can be felt! The driver in a cab of a truck or in a car or any other vehicle is keenly aware of the centrifugal force acting on a vehicle while it travels around a bend. At the kind of speeds associated with roundabouts, particularly on the continent it is clear that quite severe discomfort can be generated by the outward drainage causing lateral G-forces to gradually build up; long before danger is reached the drivers ease off because they can feel the lateral G-force starting to act. So what is going on and why do the continental designs have lower crash rates on the circulatory roadway? It is time to look at the next dynamic and that is angular momentum about a longitudinal axis.

The researchers have mostly failed to look at this although one study of dump trucks identified instability caused by driving over a bump on rough ground at a lateral angle that should have been well within stability but became unstable due to the bump. The dynamic that happens on UK roundabouts is a relatively sudden rotation about a longitudinal axis from front to rear of the trailer unit, specifically from the pin at the front to the base of the rear wheels - the axis about which rollover is initiated.

Imagine that you are following a truck circulating a UK roundabout; you will notice that it is tipping to the right perhaps up to 4. As the vehicle exits the driver will start to straighten the steering then turn to the left to exit. Once on the exit the truck may now be leaning 3-4 to left. It is this rotation that has to be achieved somehow and initially if the manoeuvre is taken at some speed there will be a tendency for the sprung part to remain at the angle it was and then to suffer angular acceleration anti-clockwise as you see it from behind. This could result in a lesser reactive load on the nearside (left) wheels and they might leave the ground for a short period. The trailer is now gaining anti-clockwise angular momentum rapidly and reaches the point when the road is now at a steady slope to the nearside say 3. But the angular momentum gained is trying to be conserved and will continue the rotation unless resisted. But the trailer is already leaning 3 in the direction that this momentum is trying to rotate it and it is now dependent upon there not being sufficient momentum to lift the offside wheels to the point that complete rollover occurs. It is known that most of the UK roundabout rollovers occur on exit.

This scenario is illustrated graphically below:...

A series of drawings to illustrate how rollover is generated from 
camber reversals. Imagine that you are following this truck on a
UK roundabout. Initially it is circulating to the RIGHT. But the driver
starts to exit LEFT and the truck rolls over the crown.
1. Truck circulating normally;
the slope here is 4 inwards.
2. Driver turns to left
Slight lurch of sprung mass
to right. I have allowed 2.
This will then impart some anti-
clockwise angular momentum
to the sprung mass helped 
by the change in camber.
3. Truck has now passed over
the crown and sprung mass
has rotated about 10-12
anti-clockwise where the exit
slopes say 4 to the left. Can
the truck stabilise at this
4. Depending on the speed of 
the truck exiting the
roundabout, the angular
momentum gained from 2. to 3.
will try to maintain that rotation
and may lift the offside
wheels completely with
possible rollover.
Will the anti-clockwise angular momentum gained be enough to lift the offside wheels? 
This may not be sufficient to roll the truck over but it will be a close thing. Clearly the crowns are an issue.
Resisting the established angular momentum at 4.depends on many factors


A recent crash on a "square" roundabout over a motorway involved an articulated truck rolling towards the inside of the roundabout. This unusual incident requires further investigation to establish the exact cause and at present no information is available on account of possible legal proceedings. However, it is worth looking at the possibilities assuming that there was no mechanical failure or natural cause.

I suggest that the scenario is the mirror image of that above; in this case the truck entered the roundabout possibly at a green traffic signal since the roundabout had been signalised. 

This means that entry speed may have been much higher than usual resulting in the same effect as above but 1. occurring on the entry curvature, 2. possibly occurring as the vehicle straightened up, 3. as the truck passed onto the circulatory roadway slope now to the right while still on slight left lock, and 4. nothing to resist the clockwise angular momentum now gained (as viewed from the rear), so unless the speed was kept down, as it might need to be, rollover could happen quite easily.

An important lesson here is that it might not be safe to signalise certain large roundabouts without checking and perhaps adjusting the roadway profiles. Particularly as this was not a circular roundabout.

A26/A275 roundabout - Lewes, East Sussex, UK

"The Scalping" A26/A275 Lewes, East Sussex
"The Scalping" A26/A275 Lewes, East Sussex
Note the outward fall on the circulatory roadway,
the conspicuous central island and the environmental art.
Note the three lane high capacity entries
and the absence of any crowns.
By sloping the circulatory roadway outwards drivers should be able to visually separate the dome of the roundabout central island from the remainder of the vista including in particular from the splitter island/central reservation.
The question of draining the circulatory roadway gets much discussion in my seminars; obviously a main worry is that of trucks overturning on roundabouts, but also that of skidding in adverse weather, including possible problems for powered two wheelers. But the truth is that drivers and riders are very aware of the side-force demands that they are making between the vehicle and carriageway and drive accordingly; problems start to arise when the carriageway suddenly changes from one slope to another, sometimes causing a considerable lurch; similar problems arise if there is an insecure load which moves suddenly as a result of or also causing a lurch. As always it is important to secure good entry deflection to ensure that drivers cannot arrive at the adverse crossfall section at excessive speed.
The advantages which are apparent to designers who accept the principle of outward drainage are considerable, not only in design, construction and maintenance, but for road safety and capacity too.  


In Summary:

  1. Traffic speeds on entry, and around or across the roundabout are reduced;

  2. Capacity is increased;

  3. There are no crowns; risk of grounding or overturning is reduced;

  4. The central island, circulatory roadway and central reservation are clearly visible and distinguishable;

  5. Drainage design and maintenance are easier;

  6. In the case of new roundabouts being built on existing road lines, the provision of outward drainage can reduce utilities costs by keeping many of the construction elements of the scheme well above existing utilities levels.

Here is a short list of sites that I know of in the UK where normal roundabouts, as opposed to small (less than 8m dia), have outward drainage.  If you know others please let me know as I want to monitor their performance. I have now seen several, but mostly they are either very old or very recently installed.
There are many of a diameter around 6-8m but few much larger than that.
The Clock-Towers at Torquay and Exeter feature.

This table contains links to websites that consider the issues of roll-over
that you may find interesting: Although not specifically dedicated to truck roll-over the pages here are of considerable interest as they cover the causes and mechanics quite well. PDF of SSFs for various vehicles. Defines SSF and illustrates the point that the SSF for many vehicles would be OK if used on smooth roads but not for off-roading where high numbers of roll-overs, usually with serious consequences, occur. More on SSF and roll-over Spectacular roll-over from a dash-cam. Note carefully that the truck concerned appears not to travel in a curve but has been slightly lurched by steering adjustments. Clearly the driver was going too fast. The trailer, as usual, loses control first, dragging the tractor over later. Study of a dump truck on an uneven site. This is very relevant as the study indicates the effect of a relatively small bump (lurch) causing roll-over when all other parameters appeared to be well within safety margins. This is a near parallel to the effect of crowns on roundabouts. Another excellent study by HSE/ Sheffield Hallam University. Note the observation of the bump in the road (p35 of the PDF, p26 of the text). This feature should have been brought out much more by the researchers. The technology for Stability Control Systems exists and could be used more on large vehicles according to Indiana Injury Lawyer Blog.

A37 Dorchester northern bypass, nr Charminster, Dorset
A250 (Tesco site) Sheerness , Kent
  Sea Street, Herne Bay, Kent
A26 (north of tunnel entrance) Lewes, E Sussex
A379 The Clock Tower, Torquay Harbour (image above)
  The Clock Tower, New North Road, Exeter
A367 Bath (about 2 miles out on the Exeter Road)
A31 A350 (I spotted this one evening on the way back from the seminar
at Chichester; It was too dark to take any photos.
A379 B3192 Churchstow, S Devon (image above)
   There are actually many such sites all over the UK. These are worth noting
as they may have been operating for a very long time without issues.

Links to other pages:

Penntraff - August 2018
Pages designed by: