What would you rather have?

[Peter Freeman]

Isn't that basically just a signalised junction with segregated left turn filter lanes? I'd be surprised if that weren't in use already.

Yes it is, except that rarely is it laid out symmetrically (3-pointed star form) to remove the perception of it being (or it was previously) a T-junction.

[jackal]

Has the following three-level junction arrangement ever been considered? Essentially use the same layout as at (e.g.) Almondsbury, but with all the right-turn slips at the same level. The crossover would be controlled by a single set of signals at which all traffic was required to proceed straight ahead.

I think the Americans have discussed this as an idea, it could work if applied to less heavily used roads.

I've often pictured this arrangement too: a poor man's stack. It would be suitable at many locations. It has the characteristic of unbalanced turning facility, as does the left-turn-slipped stackabout, but definitely has higher capacity. Unfortunately its only easy upgrade path is to a full stack, and that would have to be pre-planned in order to leave space at the mid-level.

It sounds like you're basically describing a three-level SPUI.

I'm not aware that any have been built, but the upgrade trajectory would be no more difficult than for a three-level diamond, which are widely used, especially in Texas. There the standard upgrade is to add semi-direct connectors over the top, eventually (if the full upgrade is reached) creating a stack. The flat junction itself is left in place as the connection between frontage roads. There is no need to leave space at the middle level and it would usually be wasteful to do so, as it increases structural content significantly for the sake of proofing for an upgrade that may never happen and can be put on higher levels if it does.

[Peter Freeman]

I've often pictured this arrangement too: a poor man's stack. It would be suitable at many locations. It has the characteristic of unbalanced turning facility, as does the left-turn-slipped stackabout, but definitely has higher capacity. Unfortunately its only easy upgrade path is to a full stack, and that would have to be pre-planned in order to leave space at the mid-level.

It sounds like you're basically describing a three-level SPUI.

Yes, that would be a good description. I am familiar with SPUIs: we have four in Melbourne and several more in other Australian cities. I am a fan - they work really well. Although I haven't seen a 3-level example, I had no illusion that I'd just invented something!

I'm not aware that any have been built, but the upgrade trajectory would be no more difficult than for a three-level diamond, which are widely used, especially in Texas. There the standard upgrade is to add semi-direct connectors over the top, eventually (if the full upgrade is reached) creating a stack. The flat junction itself is left in place as the connection between frontage roads. There is no need to leave space at the middle level and it would usually be wasteful to do so, as it increases structural content significantly for the sake of proofing for an upgrade that may never happen and can be put on higher levels if it does.

And yes, I understand this too: I spent some time in San Antonio a few years ago and encountered those 3-level diamonds. I think of them as square signalised stackabouts, but much better than UK circular stackabouts because they make no pretence to be roundabouts and, in my view, are safer. I'm also aware of Texas's propensity to upgrade them by progressively overlaying direct connectors, but that means that the connectors are not at the optimal (ie. middle) level for a stack.

[jackal]

I've often pictured this arrangement too: a poor man's stack. It would be suitable at many locations. It has the characteristic of unbalanced turning facility, as does the left-turn-slipped stackabout, but definitely has higher capacity. Unfortunately its only easy upgrade path is to a full stack, and that would have to be pre-planned in order to leave space at the mid-level.

It sounds like you're basically describing a three-level SPUI.

Yes, that would be a good description. I am familiar with SPUIs: we have four in Melbourne and several more in other Australian cities. I am a fan - they work really well. Although I haven't seen a 3-level example, I had no illusion that I'd just invented something!

I'm not aware that any have been built, but the upgrade trajectory would be no more difficult than for a three-level diamond, which are widely used, especially in Texas. There the standard upgrade is to add semi-direct connectors over the top, eventually (if the full upgrade is reached) creating a stack. The flat junction itself is left in place as the connection between frontage roads. There is no need to leave space at the middle level and it would usually be wasteful to do so, as it increases structural content significantly for the sake of proofing for an upgrade that may never happen and can be put on higher levels if it does.

And yes, I understand this too: I spent some time in San Antonio a few years ago and encountered those 3-level diamonds. I think of them as square signalised stackabouts, but much better than UK circular stackabouts because they make no pretence to be roundabouts and, in my view, are safer. I'm also aware of Texas's propensity to upgrade them by progressively overlaying direct connectors, but that means that the connectors are not at the optimal (ie. middle) level for a stack.

Thanks for reply. A couple of thoughts:

1) As I understand it, safety considerations actually tell in favour of roundabouts generally rather than signalized crossroads. A major consideration is the deflection and associated 'forced' reduction in speed at a roundabout. At a signalized T-junction, at least in standard configuration, there is a significant risk of a T bone or other high speed collision where a driver misses a light or where there is a malfunction. Although you have collisions at roundabouts they are generally slower with a lower risk of serious injury. How much this carries through to stackabout vs. three-level diamond vs. three-level SPUI I don't know though there would seem to be similar considerations in play.

2) It's an interesting question what the 'optimal' arrangement for a stack is. Many combinations can be found:

M4/M25, M4/M5, Madrid and Durban stacks: mainlines levels 2 and 3, connectors 1 and 4
M23/M25: the arrangement you suggested, with mainlines at 1 and 4, connectors 2 and 3
Athens and The Hague stacks: mainline levels 1 and 2, connectors 3 and 4
Sydney stack: unusual arrangement with mainlines levels 1 and 3, connectors 2 and 4.
Wetzlar stack: another unusual arrangement with one connector at level 1, a mainline at 2, a pair of connectors at 3, one connector at 4, another mainline at 5.

That should cover all stacks in Europe and the Southern Hemisphere. In Texas the tendency seems to be connectors over the top, like Athens and The Hague, even when the stack was built from scratch rather than in stages, though with many variations.

From this limited survey I'm not sure I'd say with any confidence that there is really any optimal arrangement. Probably site specific factors (such as any existing junctions or other infrastructure, the lay of the land, turning volumes, etc) tend to dominate.

[Peter Freeman]

A signalised cloverleaf would work if it has C/D roads.

Yes, that's true - if the C/D road then merged properly into the mainline without signals.

Sotonsteve - is that what you were thinking of?

It was, yes. I have personally never experienced a cloverleaf which doesn't have the turning movements segregated from the mainline. And on the subject of introducing "new" junction layouts to a country which barely uses cloverleafs, having a signal controlled cloverleaf would make the experience less daunting, ... (snip)

As noted elsewhere in this thread, much as we denigrate them, many cloverleafs still work adequately in the USA. And as Booshank and others have pointed out, many others have been fixed by conversion to clover-stacks: opposite quadrants retain their loops while two semidirect connectors provide the other two right turns. In-between sites, requiring help but unable to justify clover-stack conversion, could use C/D lane traffic signals as Sotonsteve suggested.

That would be a cheap fix for an existing but over-capacity site, removing the problem of scary merging on the C/D lanes. Only the problem merges need be signalized, not all four. I wonder whether it's been done anywhere? Definitely not something to build new now of course.

And before anyone cries out "traffic signals on a motorway?!" - why not, if they produce the desired outcome?

[Stevie D]

As noted elsewhere in this thread, much as we denigrate them, many cloverleafs still work adequately in the USA.

It depends what you mean by "work adequately". If you mean "connect two moderately trafficked roads with an inefficient use of landspace and a poor accident record" and you don't have any higher aspirations than a bad way to achieve the bare minimum then yes, they do manage that. But I have yet to come across any circumstances where a junction works better – or even as well – as a cloverleaf than in another possible configuration within the same footprint, especially when you take safety into account.

[Peter Freeman]

I'm not defending the cloverleaf to the point of suggesting we build them, or that we should not address their inadequacy where it's severe. The design was a fiendishly clever idea in the first half of last century, but we now know better. My point was that the USA (in particular) has a legacy of these things (rather like UK motorways have a legacy of inappropriate 2-bridge roundabouts), but you don't need to rip them all up willy-nilly. I've personally driven through many USA cloverleafs, even examples that don't have C/D's, and have survived. There are low-trafficked parts of that large country. I don't have accident figures, and that is a factor and a reason to fix a site that does have that problem. Each location on it's merits.

The best fix, often, is conversion to a clover-stack. That resolves all the weaving issues. In fact, a clover-stack is a viable interchange type to build today. Its defect is land-take, but I would advocate compacting it and let users slow down!

Other conversions are also often viable, since the cloverleaf's single central bridge structure is agnostic in respect of virtually all M-to-M interchange types.

[jackal]

Well you actually have a cloverstack with offside right turns there, so a two bridge layout (like a two bridge roundabout) would actually be well suited to carrying the split carriageways. See the M50 upgrades, e.g. N2:

2008

2012

The more conventional clovermill-type design (with windmill or octopus type turns to/from the nearside) also uses parallel or near parallel bridges. Here's one I sketched for an M40 J9 upgrade:

There's no difficulty in upgrading a two-bridge roundabout to a freeflow design, but there is a lack of will in the UK.

[ABB125]

What are the advantages and disadvantages of a clovermill compared with a cloverstack (without right-hand merges) and whatever the M40/M25 junction is? They are all very similar.

[KeithW]

I rather like the A19/A66 junction before the Tees Viaduct - it does seem to work remarkably well with a lot of turning traffic. Of course it helps that the Tees Viaduct was designed to be high enough for ships to pass underneath.

https://www.google.co.uk/maps/@54.56334 ... !1e3?hl=en

[Stevie D]

What are the advantages and disadvantages of a clovermill compared with a cloverstack (without right-hand merges) and whatever the M40/M25 junction is? They are all very similar.

A lot of it comes down to whatever fits the local geography best – looking at the angle the roads meet at, the vertical profile, any other roads or buildings that could fall within the footprint of some layouts.

It's generally better to avoid right-hand exits and entries, unless that turn is likely to be the dominant movement*. One interesting feature of the clovermill as shown there is that all roads have just a single exit, which then splits – whereas junctions like M25/M40 and M6/M62 have one road where both directions have a single exit and one road where both directions have separate exits for each turn. The single exit option might be simpler in terms of driver understanding, which could potentially reduce dangerous manoeuvres or drivers taking the wrong road, but looks like it could increase construction costs and landtake, and may possibly lead to more congestion at the exit if all traffic is coming off at the same place ... but TBH that's all guesswork, I haven't seen any evidence or research on it either way.

* This doesn't have to be a reciprocal arrangement. If you had one very busy road crossing a road that was a lot less busy, it could well be that the major movement for traffic on the main road was straight on, but the major movement on one arm of the minor road was a left-turn, for example.

[booshank]

As noted elsewhere in this thread, much as we denigrate them, many cloverleafs still work adequately in the USA.

It depends what you mean by "work adequately". If you mean "connect two moderately trafficked roads with an inefficient use of landspace and a poor accident record" and you don't have any higher aspirations than a bad way to achieve the bare minimum then yes, they do manage that. But I have yet to come across any circumstances where a junction works better – or even as well – as a cloverleaf than in another possible configuration within the same footprint, especially when you take safety into account.

I think you're missing the point though, which is that the cloverleaf is the most basic type of freeflow 4-way interchange. Of course there are better freeflow interchange types that typically involve successively replacing loops with semi-directional ramps all the way up to a 4-level stack, but they involve more engineering and more cost. I think the question really is whether a basic freeflow interchange (cloverleaf) is better than a non-freeflow interchange (roundabout of some sort, signalised interchange of some sort etc).

[booshank]

Locally to me an example of a junction that might be cloverleafed in other countries could be M4 J18 (A46). It's too busy for the current 2-level roundabout, hence the traffic lights, but not busy enough for a more elaborate free flow interchange.

[jackal]

What are the advantages and disadvantages of a clovermill compared with a cloverstack (without right-hand merges) and whatever the M40/M25 junction is? They are all very similar.

Yes, they basically all do the same job in slightly different ways. Supposing two of our right turns (in opposite corners) are loops, and we want to avoid weaving, there are three main types of freeflow right turn we can add:

1. Stack style - goes inside or over both loops. Add two of these right turns for a cloverstack. E.g. https://www.google.co.uk/maps/@48.9742598,2.4878214,15z
2. Turbine style - goes outside both loops. Add two of these right turns for a cloverturbine. E.g. https://www.google.co.uk/maps/@53.57406 ... 642471,14z
3. Windmill style - goes outside one loop, inside the other. Add two of these right turns for a Clovermill. https://www.google.co.uk/maps/@51.5437246,7.1992415,15z

(A fourth option is to retrofit loops with bridges to remove weaving, though it is far from optimal.)

Where does the "partially-unrolled cloverleaf" like M25/M40 fit in? It's what you get from one stack-style turn and one turbine-style turn. Other irregular combinations, e.g. stack plus windmill, are possible but less common.

As for selecting between them, site constraints and geography will be major factors as mentioned above. Still, some general points can be made. A cloverstack is usually on three levels, so is space efficient but inappropriate where visual intrusion is a concern. Cloverturbines tend to have high design speeds like stacks, with the added benefit of being on only two levels, but have high land take. A partially-unrolled cloverleaf is broadly similar but with lower structural cost and less optimal slip alignment (hence they're either very big or have reduced speed). A clovermill sacrifices speed for a compact and cost effective design.

[Peter Freeman]

Well you actually have a cloverstack with offside right turns there, ...

Oops. I inserted that image at the last moment in case any members or lurkers weren't familiar with a cloverstack but, in-haste and carelessly, I used that version without checking which one I'd used. I usually think first of the version that has semi-directional connectors that pass over the remaining loops, but still within all the left turn connectors. More bridges required.

However, I must say that I do like the one I used: directional connectors, right turn exits on the offside. All the usual caveats about that of course, but it's neat, compact, structurally economical. Fitting the ramps in the median might be tight in some cases. I've not noticed any of these in real life - they exist?

[Peter Freeman]

There's no difficulty in upgrading a two-bridge roundabout to a freeflow design, but there is a lack of will in the UK.

Yes, as long as the bridges are parallel-ish and straight-ish. And as long as the site is not small and closely surrounded by development.

What is really difficult is conversion of a stackabout to a different M-to-M (system) interchange. Most UK stackabouts are circular, with curved roundabout bridges: hard to re-use.

[jackal]

Well you actually have a cloverstack with offside right turns there, ...

Oops. I inserted that image at the last moment in case any members or lurkers weren't familiar with a cloverstack but, in-haste and carelessly, I used that version without checking which one I'd used. I usually think first of the version that has semi-directional connectors that pass over the remaining loops, but still within all the left turn connectors. More bridges required.

However, I must say that I do like the one I used: directional connectors, right turn exits on the offside. All the usual caveats about that of course, but it's neat, compact, structurally economical. Fitting the ramps in the median might be tight in some cases. I've not noticed any of these in real life - they exist?

You really have to pick one of 'compact' and 'structurally economical' for this design - either you spread the carriageways for earthworks or (as in your image) you require long structures to gain height for the direct right turns. And even in the former case it's still a three level design so not that cheap. The offside clovermill-type design (like the M50/N2 image above) achieves much the same but at lower cost.

This in Oklahoma is the closest I've seen to your image but the direct ramps merge and diverge from the offside:

https://www.google.co.uk/maps/@35.59977 ... 288087,16z

[booshank]

What are the advantages and disadvantages of a clovermill compared with a cloverstack (without right-hand merges) and whatever the M40/M25 junction is? They are all very similar.

Yes, they basically all do the same job in slightly different ways. Supposing two of our right turns (in opposite corners) are loops, and we want to avoid weaving, there are three main types of freeflow right turn we can add:

1. Stack style - goes inside or over both loops. Add two of these right turns for a cloverstack. E.g. https://www.google.co.uk/maps/@48.9742598,2.4878214,15z
2. Turbine style - goes outside both loops. Add two of these right turns for a cloverturbine. E.g. https://www.google.co.uk/maps/@53.57406 ... 642471,14z
3. Windmill style - goes outside one loop, inside the other. Add two of these right turns for a Clovermill. https://www.google.co.uk/maps/@51.5437246,7.1992415,15z

(A fourth option is to retrofit loops with bridges to remove weaving, though it is far from optimal.)

Where does the "partially-unrolled cloverleaf" like M25/M40 fit in? It's what you get from one stack-style turn and one turbine-style turn. Other irregular combinations, e.g. stack plus windmill, are possible but less common.

As for selecting between them, site constraints and geography will be major factors as mentioned above. Still, some general points can be made. A cloverstack is usually on three levels, so is space efficient but inappropriate where visual intrusion is a concern. Cloverturbines tend to have high design speeds like stacks, with the added benefit of being on only two levels, but have high land take. A partially-unrolled cloverleaf is broadly similar but with lower structural cost and less optimal slip alignment (hence they're either very big or have reduced speed). A clovermill sacrifices speed for a compact and cost effective design.

Interesting! Aren't the main determinants of size though the size of the loops and radius of the left turn movements (right turn in RHD), which in turn depends on the angle at which the mainlines cross?

For example the Paris cloverstack example is only relatively compact because the mainlines don't cross at right angles and so two of the right turn movements can be brought in towards the crossover point. If they crossed at right angles, you'd probably want them further out to give a decent radius, which would leave space for the two extra loops of a full cloverleaf, even if those loops weren't there.

[jackal]

By space efficient I meant something like 'land take relative to design speed'. A three-level cloverstack is space efficient as it has the absolute minimum land take in two corners (just the short turns). So yes, higher radius turns increase land take, but it will still use less space than other designs using similar radii.

[booshank]

By space efficient I meant something like 'land take relative to design speed'. A three-level cloverstack is space efficient as it has the absolute minimum land take in two corners (just the short turns). So yes, higher radius turns increase land take, but it will still use less space than other designs using similar radii.

But will it? Surely if a cloverstack is going to have the same design speed as a full cloverleaf, the analogous left turns will be the same radius and therefore the footprint of the interchange will be the same. The only difference will be one or more loops are replaced with semi-directional ramps, which doesn't change the footprint. Or am I missing something?