On-ramps at empty-terrain diamond interchanges: should they have S-curves?

[J N Winkler]

My motivation for starting this thread is to get other people's views on a problem in the geometric design of freeway on-ramps I have been thinking about for a while.  The "empty terrain" and "diamond interchange" conditions in the thread title are designed to get at the essentials by ruling out other factors which could influence on-ramp design, such as the presence of building development or loop ramps between the on-ramp and the freeway mainline.  For purposes of discussion it is assumed that the freeway remains straight through the interchange and that the spaces left between the ramp terminals and the crossing point between the surface road and freeway are essentially arbitrary (i.e., governed by engineering and planning policy considerations unrelated to this problem).

The purpose of a freeway on-ramp is to allow a gradual increase in speed between the start of the ramp and the merging area.  To avoid building retaining walls between the freeway mainline and the ramp, it is generally necessary for the start of the ramp to be some distance away from the point where the freeway mainline and the surface road cross each other, and therefore there has to be at least one curve on the ramp between its starting point and the merging area.  The purpose of the curve or curves that are provided is to change the bearing of ramp traffic to match that of traffic on the freeway mainline.

The simplest solution is to have just one curve, whose beginning point is some distance upstream of the merge nose and whose trailing transition ends just downstream of the nose.  The problem with this approach is that the curve is so close to the merging point that it has to be negotiated at close to the operating speed of the freeway mainline, and it also has to handle a large change in bearing (delta) which is essentially the same as the angle between the ramp and the freeway mainline.  Unless the curve is made quite long, which can have implications for right-of-way acquisition, side friction demand going through it can rise to levels which tend to throttle the speeds of merging cars at precisely the moment when they need that speed to merge safely.  Another issue is that the ramp does not form a right-angle intersection with the surface street, which can promote easy right turns and difficult left turns unless channelization is provided to ease the left turns.

A more elaborate solution is to provide a reverse curve between the start of the ramp and the merging area, with both curves possibly having tighter radii and less superelevation than a single curve located near the merging area, but being designed to be negotiated at much smaller speeds.  Vehicles come out of these curves with some speed (the amount depends on the location and geometric characteristics of the curves), well upstream of the merging area, and with ramp crossfall and bearing more or less matched to that of the freeway mainline.  This gives drivers more room to match speeds on the freeway, eliminates the throttling effect of having a single curve so close to the merging area, and reduces driver task loading by eliminating the need to negotiate a curve and search for a gap in traffic simultaneously.

My questions:  is the more elaborate solution generally worthwhile?  If it is not, in the general case, then what are the conditions under which it becomes better than the simple one?

In the real world, the simple solution seems to be preferred by Kansas DOT and other Midwestern state DOTs while the S-curve approach seems to be a coastal state DOT (Caltrans, et al.) specialty.  (I welcome any exceptions to these generalizations.)

[froggie]

The S-curve approach is also very much an LaDOTD thing too.

IMO, the "elaborate" solution may be more appropriate for off-ramps where the goal is to slow traffic down before it gets to the end of the ramp.  But for on-ramps, unless there's another underlying reason have an S-curve (adjacent development, wetlands, etc etc), they shouldn't be used...ideally, on-ramp traffic should be merging onto the freeway at highway speed, and having additional curves to negotiate detracts from that.

[agentsteel53]

it might help to look at the problem explicitly from the perspective of: what benefit does each curve provide?

the second curve - closer to the merge point, allows a ramp which is coming at an oblique direction to be straightened out gradually to meet the traffic much closer to parallel.

the first curve - given that the second curve is in place, extending the ramp backwards straight from the point of intersection would cause the ramp to have a combination of easy right, difficult left.  to restore balance between the two directions, the second curve is placed, allowing the ramp to depart the side street at a 90 degree angle.  Also, space is saved by having the entry point be placed closer to the freeway.

so, essentially, I would use the S curve where right and left turns were about equally in demand and there was no place for a Y-merge (extra channel) for left and right traffic onto the on-ramp*.

if there is a bias towards rightward-turning traffic, then the easier right/more difficult left ramp may very well still be suitable.

generally, though, I'd expect enough symmetry between left- and right-merging traffic to justify the S. 

[agentsteel53]

also, some diagrams, JNW?    

I think we're on the same page with our verbal descriptions but it would be helpful to confirm.  If I have the time tonight, I can come up with some diagrams to illustrate what I was just describing.

[J N Winkler]

Jake--I think we are on the same page, though I am not sure that what happens near the beginning of the ramp happens at a high enough speed to be the determining factor for a choice between the "single curve" and "double curve" solutions.

Here is an example of the "single curve" solution:

http://maps.google.com/maps?q=Maize,+Kansas&ie=UTF8&hq=&hnear=Maize,+Sedgwick,+Kansas&ll=37.77911,-97.459502&spn=0.004316,0.009645&t=h&z=17

I have taken this ramp a number of times--it is long enough between the cross road and the merge nose that you can get up to 70 MPH easily, but then you have to let up on the throttle and slew hard to the right to get past the merge nose.

"Two-curve" solution (fairly subtle in this case):

http://maps.google.com/maps?q=Baton+Rouge,+Louisiana&ie=UTF8&hq=&hnear=Baton+Rouge,+East+Baton+Rouge,+Louisiana&ll=30.418827,-91.431168&spn=0.002355,0.004823&t=h&z=18

[agentsteel53]

Here is an example of the "single curve" solution:

that is an example which I had not thought about - where the final merge features an angular difference (between the on-ramp and 96 south) which is sufficiently large that the driver has to actively correct for it. 

this in opposition to the Louisiana example that you also posted, in which the final angle at the end of the on-ramp is small.

I would define "small" to be a smaller angle than an ordinary lane-changing between parallel lanes by a driver doing so at a reasonable pace.  if the on-ramp allows you to merge at that pace, then it is small - if it requires a greater turn of the steering wheel, it is large.

so maybe we need to separate the situation into "small" vs "large" final approaches.  One thing that a well-designed two-curve solution can do is make the final merge angle "small". 

[Revive 755]

"Two-curve" solution (fairly subtle in this case):

http://maps.google.com/maps?q=Baton+Rouge,+Louisiana&ie=UTF8&hq=&hnear=Baton+Rouge,+East+Baton+Rouge,+Louisiana&ll=30.418827,-91.431168&spn=0.002355,0.004823&t=h&z=18

I think this one on I-70 in Illinois is a much better example:
http://maps.google.com/maps?hl=en&ie=UTF8&ll=38.786572,-89.626358&spn=0.013732,0.033023&t=h&z=16

Indiana and Ohio also seem to use the two curve design.

I'm thinking there may be an issue with land required for a single curve design in which a large spacing between ramp terminals on the cross road is desired.  Since this seems to very somewhat by state, it probably helps influence the choice of design.

[Duke87]

"Two-curve" solution

I think this one on I-70 in Illinois is a much better example:
http://maps.google.com/maps?hl=en&ie=UTF8&ll=38.786572,-89.626358&spn=0.013732,0.033023&t=h&z=16

Yeah, that's more what I was thinking was meant. This sort of design really seems to be a waste of land.

[ftballfan]

The US-31 freeway in Oceana County is infamous for its land-hogging diamonds. Heck, the thirteen ramp interchange at US-31 and I-96 is smaller than some of those diamonds.

[roadfro]

It's kind of a crapshoot when it comes to Nevada. There's a mix, but on casual observation it seems that NDOT prefers the S-curve slightly over the single-curve design in the empty/rural setting...likely for the reasons that JN Winkler mentioned. I'm not sure about any actual design policy that NDOT has (or if the AASHTO Green Book) has anything on this)...

[mightyace]

Both exits along Saturn Parkway have S-curves.

http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=Spring+Hill,+TN&aq=1&sll=37.0625,-95.677068&sspn=31.28862,78.134766&ie=UTF8&hq=&hnear=Spring+Hill,+Maury,+Tennessee&ll=35.733763,-86.904345&spn=0.031283,0.076303&t=h&z=14

Would this be considered two or three curves on these ramps?

Now, some erroneous local maps show the Port Royal Rd interchange as a cloverleaf as well as cloverleaf loop ramps on the south side (Eastbound) of the Kedron Rd. interchange.  Both on the map and driving them, it certainly looks like tight loops could have been built, but I'm glad they weren't.