Roundabout capacity: entry width, flare and the inputs

A roundabout capacity assessment is mostly an exercise in measuring six pieces of geometry correctly. The empirical method used across the United Kingdom derives entry capacity from the physical shape of each approach, and small errors in those measurements move the answer more than most people assume. A flare measured from the wrong point can turn a junction that fails into one that works, on paper.

The method comes from Transport Research Laboratory work on entry capacity at roundabouts, which produced a linear relationship between entry capacity and circulating flow, with the slope and intercept set by geometry. Every standard piece of roundabout modelling software in use here implements a version of it, which is why a highway officer can check your inputs without owning your software.

The geometric parameters that drive capacity

Six measurements do the work on each arm.

• Entry width, measured at the give-way line perpendicular to the nearside kerb.
• Approach half-width, measured upstream of any flare, from the kerb to the centre line or to the median.
• Effective flare length, which describes how the approach widens into the entry.
• Entry radius, the radius of the nearside kerb on entry.
• Entry angle, the angle between the entry path and the circulating path.
• Inscribed circle diameter, the largest circle that fits within the junction outline.

Entry width and flare length dominate. A wide entry with a long flare accepts more vehicles per unit of circulating flow, because more drivers can present themselves at the give-way line at once. Inscribed circle diameter has a smaller effect and entry angle smaller still, although at extreme values both matter.

The important point about the method is that the geometry is measured from the built layout, not from the drawing conventions. What you enter should describe what a driver actually meets.

Measuring entry width and effective flare

Entry width is measured at the give-way line, perpendicular to the nearside kerb, from that kerb to the point where the entry meets the circulatory carriageway. Where there is a splitter island the measurement stops at the island, not at the centre of the road.

Effective flare length is the parameter most often got wrong. It is not the physical length of the widened section. It is derived from the distance over which the entry widens, measured along a line drawn at a defined offset from the nearside kerb, and it reflects how much of the widening a driver can actually use. A long taper that only reaches full width at the give-way line delivers far less capacity than its physical length suggests.

Two habits produce defensible inputs. Measure from an accurate topographic survey rather than from mapping, because kerb lines on background mapping are approximate and a 0.3 metre error in entry width is material. And photograph what you measured, with the measurement marked, so the highway officer can see the basis.

For a proposed junction rather than an existing one, take the geometry from the general arrangement drawing and state the drawing number and revision in the assessment. If the layout changes later, the model has to be rerun, and a stated revision number makes that obvious rather than hidden.

Interpreting the ratio of flow to capacity

The output that matters most is the ratio of flow to capacity, usually written as RFC. It is the demand flow on an arm divided by the capacity of that arm, both in passenger car units per hour, for a given time segment.

The conventional practical limit applied by highway authorities in the United Kingdom is 0.85. Below that, queues stay short and stable. Above it, small increases in demand produce large increases in queue and delay, because the relationship is not linear near capacity. Authorities set their own expectations and some accept higher values on urban arms where queuing is already normal, so confirm the threshold with the highway officer rather than assuming.

Read the value alongside the time segment it belongs to. Most assessments model a peak hour divided into segments, and the RFC quoted should be the worst segment rather than the hour average. An arm that averages 0.82 across the hour and reaches 0.97 in the middle segment is not a comfortable arm.

Queue and delay outputs

Capacity models report a mean maximum queue, which is the average across modelled peak periods of the longest queue reached in each. It is not the longest queue that will ever occur, and presenting it as such invites a correction.

Queue length is where a model meets physical reality. A predicted queue of 14 passenger car units on an arm with 60 metres to the previous junction is a blocking problem regardless of what the RFC says. Check every queue output against the available storage on that approach, and say in the text where the storage is.

Delay outputs are per vehicle and in seconds, and they matter most for comparing options. A scheme that adds three seconds of average delay is presenting a different case from one that adds forty, even where both stay below the RFC threshold.

Sensitivity to small geometry changes

Because capacity is derived from geometry, a modest layout change can move the answer significantly. Widening an entry by half a metre, extending a flare, or tightening the inscribed circle diameter each shift capacity.

This cuts both ways. It is a legitimate design tool, and an arm that fails at 0.91 can often be brought within the threshold by an achievable widening. It is also a temptation, and highway officers know it. An assessment that reaches an acceptable answer only with geometry that cannot be built within the highway boundary, or that conflicts with visibility splays or pedestrian crossing provision, will not survive review.

Run and report the sensitivity. Showing how the result moves with a half-metre change in entry width tells the officer that the design has margin, or that it does not. Sensitivity testing is also the honest answer to uncertainty in the traffic inputs, which is usually larger than the uncertainty in the geometry.

One further caution. Empirical roundabout methods were developed for conventional geometry and behave less reliably on very small roundabouts, on grade-separated layouts and on junctions with heavy pedestrian and cycle demand across the arms. Where the layout departs from the norm, say so and consider whether a different approach is needed.

Presenting results to a highway officer

The reviewer wants to check your work, not admire it. Give them what they need in the order they will look for it.

1. A plan showing each arm numbered, with the measured geometry annotated on it.
2. A table of geometric inputs by arm, with the source of each measurement stated.
3. The traffic flows used, with their derivation and growth assumptions, and the assessment years.
4. Results by arm and by time segment: RFC, mean maximum queue and average delay.
5. Available queue storage on each approach, compared against the queue outputs.
6. Any mitigation modelled, with the geometry it assumes and whether it fits the highway boundary.

Keep the model output files available and offer them. Officers who can run your model themselves resolve queries in a day rather than in a round of correspondence.

The capacity work sits within the wider evidence base, and on a scheme large enough to need a Transport Assessment the junction results support the access case rather than standing alone. On smaller schemes a Transport Statement may carry a single junction assessment, and the same presentation discipline applies. Where the junction lies on or near the strategic road network the modelling will be reviewed more closely still, and the position set out in when National Highways objects to an application is worth understanding before submission.

Capacity assessment is work for a suitably qualified transport planner, and councils expect that. PlanScope's drafts support the written submission rather than the modelling itself.

This is general guidance, not legal advice. Check the position with your local highway authority or a qualified professional before relying on it.

Enter your postcode on PlanScope to see which transport documents your scheme is likely to need.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.