DMRB TD 54/07 Design of Mini-roundabouts

My comments on TD 54/07

This document (and others) replaces TD 16/93. Considering the long time it has taken since 1993 this document should have been much better than it is. The emphasis on deflection has become seriously blurred resulting in advice not to provide the full deflection recommended for normal roundabouts.

A review of such an important document should be able to cover all aspects including legislation but the definition of the central island of a mini-roundabout was not changed and should have been. I hope it does not take another 14 years to get that corrected in the UK. 

If you are reading this from outside the UK,
Do not make the mistake of setting a maximum central island diameter in your guidelines.

There is a serious failure to differentiate between three types of entry/circulating crashes that involve merging, crossing and, at 4-arm sites, right angle crossing. You need to understand the processes that lead to these crashes.

Also the data is taken from the TRL Report 281 which drew on crash data from 1986-1992 well before any serious advice on mini-roundabout design and use was available. Although useful, it begs the question as to what accident patterns would emerge were there to be a series of mini-roundabouts all designed to good basic principles outlined here and in my publications.


Red Text (this column) is compulsory on trunk roads
My comment Action
1.1 Central islands 4m wide max. I suppose I could hardly expect HA to recommend changes here despite my protests over many years. This would hang on changing the UK TSRGD, so this error and all the problems associated with too small central islands at mini-roundabouts will continue for another decade or more.

DfT and HA must realise that deflection at mini-roundabouts can only be achieved with larger central islands in many instances. Yet in both TD 54/07 and Mini roundabouts good practice guidance we are recommended not to use the central island as a speed reducing device! Absurd!

Highway authorities must learn to use common sense here and apply real deflection criteria based on forced vehicle path curvature or vertical deflections.

DfT and HA must "get real" on this issue. Watch any reasonably well installed mini-roundabout with a raised centre and most drivers will be deflected by it.

1.10 Mini-roundabouts with standard advisory give way markings rely on the "priority to thy right" rule for traffic (diagram 611.1). Not so. The sole function of sign 611.1 is to replace the sign to diagram 606 turn left, which cannot be provided on a mini-roundabout; its legal requirement is that it is placed with the mini-roundabout marking and drivers must pass to the left of that marking unless it is not possible to do so. The give-way rule remains, correctly, advisory unless a mandatory form of give-way marking is used.  
2.1 Mini-roundabouts must only be used on roads with a speed limit of 30mph... Ideally, yes, but the actual speeds are far more relevant. Remove the speed limit from the mandatory requirement.
2.15 Four arm mini-roundabouts... are not recommended where the sum of the maximum peak hour entry flows for all arms exceeds 500 veh/hr. Hardly an issue, but some busy mini-roundabouts at crossroads are difficult to drive. The larger central islands that I recommend at such sites can be very helpful at separating out the conflicts and controlling speeds.

This is in preference to reducing the ICD by adding overrun areas on the outside of the junction areas and reducing the corner radii that I have seen recently at a number of sites.

The TD should be recommending trialling these larger central island mini-roundabouts.
2.16 Mini-roundabouts with five or more arms must not be used. At such sites, before rejecting a mini-roundabout, consider one with a central island larger than 4m. One came into a seminar and was later introduced at Eldwick in the North. It works fine.
See this location in PixPlot Roundabout database. (Opens new window.)
The TD should be recommending trialling these larger central island mini-roundabouts.
2.24 The capacity at an intermediate give-way line between double mini-roundabouts will be reduced by the effect of the first junction. I do not believe that the capacity will be affected but the flow will. In most instances the amount of yielding going on will be far greater at the first yield lines than the intermediate ones. At the latter, yielding will be to right-turning (UK & LHD countries) vehicles which will usually be fewer in number than the "ahead/thru" movements which take priority at the first yield lines.  
2.3 It may be inappropriate to install mini-roundabouts on approaches to ports, industrial areas etc. No evidence - this statement should not have been made! The whole idea was that mini-roundabouts brought roundabout operation to urban sites by being overrunnable! Ignore this comment! Ensure construction is strong enough for the traffic likely to use it!
2.5 Mini-roundabouts shall not be used on dual carriageways. Mini-roundabouts are extremely successful on single lane dualling schemes such as Shenley Road, Borehamwood. According to HA single lane dualling is not a standard cross-section. Amend TD to allow this exception.
3.3 Safety - TRL accident study ... covers accidents from 1986-1992. The relevance of this point is that there was virtually no comprehensive design guidance before 1996 so the accidents all occurred at sites with hugely variable geometry and layouts. TD 16/93 was published after this period but had very little aimed specifically at mini-roundabouts anyway. It might well be appropriate to inaugurate a new safety study based on similar numbers of sites but with specific criteria built into the designs.
3.9 Accident Categories There is just one shown for entering/circulating. This is misleading. TRL report 281 has three categories:

1. Entering (merging); i.e. entering vehicle taking the next immediate exit and striking a circulating vehicle leaving at the same exit;

2. Entering (crossing); the two vehicles in conflict were crossing one anotherís paths but one would have been physically turning so only one vehicle could be moving at relatively high speed; e.g. at a T-junction mini-roundabout configuration;

3. Entering (right angle crossing - only at 4-arms) i.e. both vehicles crossing one anotherís path with potential high speed for both movements Ė this can normally only occur at a crossroads layout, but represented nearly half of the injury accidents reported at crossroads mini-roundabouts. Right angle crossing + other crossing accidents represented well over half of all injury accidents at four-arm mini-roundabouts in the study.

The techniques for dealing with these are a little different.

The document should identify these three accident types as they are significantly different and the design aspects to deal with these crash types also are different.
6.4 As vehicle speeds should already be low, full deflection as required for standard roundabouts need not be provided. I have always applied forced vehicle path curvature to crossing streams and believe that this remains a most effective way of reducing the risk and severity of crossing accidents.  "Full" deflection may include movements running along the kerb which concern me less at urban mini-roundabouts. Provided that a good approach layout is used a straight approach for the merging stream usually does not matter. But 60m radius vehicle path or less remains essential for applying to crossing streams.  A thorough revision of this section of the TD is essential.
Fig 6.2 Vehicle path diagrams

I am pleased that at last there is reference to designing mini-roundabouts on the basis of vehicle paths, which most of us have been doing since Frank Blackmore first did his "doodles" as he called those early sketches. But the opportunity has been missed to add the necessary radius to these paths to indicate that there is forced curvature; particularly necessary on the crossing movements.

6.16 Visibility.  Road users approaching the give-way line on any approach to a mini-roundabout need to be sure that it is safe to enter the circulating area. OK so far as it goes, but the first requirement is for drivers to be able to identify early enough that they are approaching a junction at which they must be expected to give-way/yield. This message must never fail to reach the driver and sight of the junction area and the approach layout are absolutely crucial. If approach speeds are higher than desirable then it is necessary for the transition area into the roundabout to be sufficiently long to allow time for adjustment. Forward visibility is part of this as set out in Fig 6.5. But the most important aspect is the appearance of the layout including the approach layout.  
Table 6/1 Visibility to the right on entry If the vehicle entering on the right is forced to 60m radius in accordance with my standard then the visibility distance becomes much less critical. Fig 6/5 is weak in that there is little or nothing done on the approach and inadequate deflection. DO NOT REPLICATE THIS LAYOUT
6.19 The give-way marking requires road users to give way to circulating traffic... In the UK the rule has always been advisory. Our layouts should reflect this more. Drivers get too upset when absolute priority is not afforded at a roundabout; but this was never intended. Shared space methodology suggests that a lot of give and take should be the way that roundabouts operate, not absolutes.  Ensure all crossing streams are well constrained.
6.24 Solid or raised areas of markings are not permitted other than for the white circle. I believe that this is wrong. I am not sure of the legislative basis but I know of many sites that use raised splitter islands very successfully. I believe that this slipped in somewhere without any real consultation. The Americans are using raised splitters but they design them somewhat differently. See Dimondale Amend the legislation on an experimental basis
6.25 A kerbed splitter island must be provided where without it, vehicles would encounter an easier path if they were to pass on the wrong side of the white circle. There may not be room for it and of itself it is not guaranteed to be effective. But an additional regulation would be broken by a driver passing to the right of a keep left sign as has been observed occasionally.  
6.33 No more than one lane must be marked as being for a given exit arm. Where is the evidence for this? I have seen some mini-roundabouts that have two lanes marked for the same exit arm, but the exit arm MUST be capable or receiving the flow; a single lane exit will not work. The degree of taper with a "merge in turn" will depend upon the level of demand on the exit.  
6.36 Typically the give-way line is placed on the circumference of the largest circle that can be inscribed within the junction kerbs. NO! Design in accordance with realistic vehicle paths. See Drawings page for an explanation why this can be all wrong especially at very small sites. This is so wrong and must be amended.
6.6 Examples of island functions These layouts are dangerous! DO NOT REPLICATE THESE DANGEROUS LAYOUTS
7 Conspicuity Note particularly 7.2 - visual aspects of the approach layout which must ensure beyond any doubt that drivers are approaching a mini-roundabout.  
7.9 Vertical features on a build-out I am concerned about build-outs but I recommend that vertical features on them should have a distinctive 3D shape and not just be vertical. e.g. shrubs and soft planting.
A raised overrunnable area might be effective - I have seen them used but not at a mini-roundabout.

Links to other pages:

(This site is subject to continual development with new pages added in February 1999, December 2000, May & October 2002 and several updates throughout 2004, 2005. MIDI-roundabouts site added 2006.)

Penntraff & MoorWeb are part of Moor Value Ltd. UK Reg. No. 5018871.

© Penntraff - Dec 2014