A multi-part article in California Highways and Public Works on freeway interchange design, the final part of which appeared in the March-April 1951 issue, prompted me to think about how rural diamond interchanges in Kansas are designed.
The classic approach, which is generally followed in level terrain and in the absence of other developed infrastructure which constrains the design, is to have the ramps leave the freeway mainline and swing out wide on tangent to termini on the surface road which are situated at a considerable distance from the grade separation between the freeway and the surface road.
This approach affords several advantages:
* Since the terminus and freeway connection for each ramp are at nearly the same level, sightlines are quite good for merging in both "freeway over" and "freeway under" situations, and the relatively modest amount of earthworks the ramp requires is more or less independent of the earthworks for the crossing structure.
* In "freeway under" situations, adequate sight distances around the bridge rail and roadside safety hardware can be provided at the termini of exit ramps without the need to provide added square footage for a shoulder on the bridge.
But there is an important disadvantage:
* For each ramp, the curve connecting it to the freeway is the principal constraint on the driver's speed. The driver encounters it early when he is leaving the freeway, thus limiting the opportunity he has to slow down after exiting, and he encounters it late when he is merging onto the freeway, which means that for purposes of getting up to speed, the generous amount of tangent alignment on the ramp is largely wasted. Where curve design is concerned, there is a considerable amount of empirical evidence that drivers will accept fairly high rates of change in lateral acceleration (up to 10 ft/sec³) provided that the change in bearing associated with each curve is very small. But since the ramp termini in these diamond interchanges are so far from the crossing structure in the middle of the interchange, the change in bearing associated with each ramp curve is quite large, which makes it necessary to adopt a conservative value for rate of change in lateral acceleration. In principle superelevation and curve radius can be manipulated to reduce side friction demand at high speeds, but this does nothing to address the awkward position of the curves near the diverge and merge noses, and can bring right-of-way constraints into play.
Is there a clever way of redistributing the curvature so that the slowest speeds are confronted with the greatest curvature and drivers can make the largest changes in speed on the straightest alignments?
I find curvature near the merge point to be much more frustrating than curvature near the exit point. Maybe this is because speed advisory signs exist upon exiting but not upon merging; also, I have the general expectation to slow down when exiting, whereas I expect to be able to be able to use the entire straightaway of a diamond interchange's on-ramp to accelerate up to highway speed.