In article m, Martin
Underwood writes
Yes, the Magic Roundabout in Swindon is a pain in the bum: it's as if the
road designers decided to make it as tortuous as possible - being cynical, I
wonder if they decided to make it hazardous so as to keep the traffic speed
down: which is silly because the deliberate hazards distract the drivers'
attention from the hazards that they should be looking for - other road
users!

And then there's the roundabout in Hemel Hempstead. This started out as one
big 6-way roundabout. It worked fairly well.
[...]

Actually, if you study both these junctions and other places where ring
junctions (to give them the proper name) are installed (e.g. the
A13/A130 intersection), you'll see that the junction was trying to
handle too much traffic and snarling up.

The basic point behind a ring junction is to reduce the average
proportion of the roundabout that traffic has to go on, thus increasing
the throughput of the junction.

Consider a ring junction with 5 exits and assume that all 20 possible
flows see equal traffic (this is to aid the explanation; at a real site
you would of course take measurements). If you build it as a normal
roundabout, the average distance that a car travels around the
roundabout is half its circumference [1]. Thus the flow on the
roundabout has to be 2.5 times the flow coming in from each road [2].

If you replace it with a ring junction, the average distance a car
travels drops to 30% of the circumference [3] and the flow only has to
be 1.5 times the incoming flows [4]. Put another way, you gain 66%
capacity (though of course you then lose some because of the additional
needs to give way, but it's still a net win).

[1] 25% of the traffic goes 20% of the circumference, 25% goes 40%, 25%
goes 60%, and 25% goes 80%. That works out as an average of 50%.
[2] The section from road 4 to road 5 carries all the traffic entering
at road 4, 75% of that entering at road 3, 50% of that entering at road
2, and 25% of that entering at road 1. That's 100+75+50+25 = 250% of the
traffic entering at any one road.
[3] Now 25% goes 20% of the circumference clockwise, 25% goes 40% of the
circumference clockwise, 25% goes 20% anticlockwise, and 25% goes 40%
anticlockwise. That's an average of 30%.
[4] The section from road 4 to road 5 now carries (clockwise) 50% of the
traffic from road 4 and 25% of that from road 3, plus (anticlockwise)
50% of that from road 5 and 25% of that from road 1. Total 150%.

Moreover,
if you are turning right, you go clockwise round each mini roundabout but
*anti-clockwise* round the central roundabout, which feels very wrong:

Live with it. Any unfamiliar layout feels wrong; what do you think about
slip roads on the right instead of the left?

--
Clive D.W. Feather, writing for himself | Home:
Tel: +44 20 8371 1138 (work) | Web: http://www.davros.org
Fax: +44 870 051 9937 | Work:
Written on my laptop; please observe the Reply-To address