Why does everyone love pairing by use, for 4-track railways

[jscz]

Having experienced the southern commuter belt of both the MML and the ECML, I am massively in favour of quadrupling by direction. However it seems that places me in the minority. The reason I think this is simple: what is fast/slow in one place might not be in another. For example, on the MML, all Thameslink trains have to use the slow tracks, no matter where they are stopping or passing. This means that the capacity for slow trains is greatly reduced while the fast tracks sit fairly empty (EMR only have about 6 trains per hour in each direction out of St. Pancras). This also means that a broken rail (or similar issue) is a nightmare as crossing between the fast and slow tracks is much more inconvenient (which is why the fast Thameslink trains don't do it).

In contrast, on the ECML, fast Thameslink trains can hop between the fast and slow lines with ease, allowing them to switch based on their stopping patterns. For example, the Cambridge-Brighton service joins the ECML on the slow tracks, then after the Digswell viaduct when it expands to four tracks again it stays on the fast lines. Then, at Finsbury Park it crosses over to the semi-fast platform, which then takes it on the slow track towards the canal tunnels. Equally, when there was a broken rail on the up fast a couple of months ago, it was fairly easy for fast trains to cross over to the down fast at the crossing before, and back at the crossing after. Had that been on the southern MML, all hell would break loose.

The other final advantage is based on the fact that realistically, all four-track lines go down to two at some point, whether that's because the line was never quadrupled that far, or because of infrastructural reasons. With quadrupling by direction, this is much more convenient than trying to do this with distance quadrupling, because trains going from one set of tracks to the other don't have to cross another line in the process.

Equally, converging and diverging lines are easier as with directional quadrupling, you can use the following layout:

While on distance quadrupling, you have to do this:

(note while the lack of a flyover/under here makes it cheaper, that's only because building this junction graded is effectively impossible)

If you disagree with my assertions, or agree but would like to play devil's advocate, please feel free to do so . I am happy to be wrong as long as I can see why I'm wrong

 

[Tomnick]

Fast Thameslink trains *do* cross from fast to slow and vice-versa on the Midland, though. They usually run fast line from West Hampstead (ish) to Radlett or Harpenden, overtaking a slower one in the process. Yes, it makes timetabling more intricate, as it needs a gap in the traffic on the line that it's crossing, but the trade-off for that is that it makes turnbacks at St Albans, Luton and Bedford much easier without having conflicting moves or expensive flyovers, as well as making it easier to access freight facilities which are all on the slow line side.

The actual reason, though, is really no more than that's how the Midland Railway built it, with a focus on keeping freight traffic separate to passenger, whereas the GN evolved with more of a focus on suburban passenger traffic.

 

[MarlowDonkey]

Having experienced the southern commuter belt of both the MML and the ECML, I am massively in favour of quadrupling by direction. However it seems that places me in the minority. The reason I think this is simple: what is fast/slow in one place might not be in another.

The GWR line out of Paddington pairs by use classifying the faster lines as Main and the other two as reliefs. Not all stations have all 4 platforms. The Reliefs are to the north of the Mains and so are most of the branch lines and freight yards. At Airport Junction, the double track to the Airport connects to all four lines.

By contrast the Tfl line from Finchley Road to Wembley Park runs fast (Met), slow (Jubilee) * 2, fast (Met). However between Baron's Court and Acton Town the line runs slow (District), fast (Pic) * 2, slow (District).

I'm not totally familiar with how the WCML operates other than noting that the DC lines are paired by use, even though I think the WCML proper is paired by direction.

 

[edwin_m]

The GWR line out of Paddington pairs by use classifying the faster lines as Main and the other two as reliefs. Not all stations have all 4 platforms. The Reliefs are to the north of the Mains and so are most of the branch lines and freight yards.

By contrast the Tfl line from Finchley Road to Wembley Park runs fast (Met), slow (Jubilee) * 2, fast (Met). However between Baron's Court and Acton Town the line runs slow (District), fast (Pic) * 2, slow (District).

I'm not totally familiar with how the WCML operates other than noting that the DC lines are paired by use, even though I think the WCML proper is paired by direction.

The WCML is paired by use from Euston to Rugby (the Slows forming the Northampton loop). On to Stafford it's paired by direction, then back to pairing by use as far as Crewe, then by direction again.

Usually with pairing by direction the fast lines are in the middle, the Met/Jubilee being a rare exception (which also has two separate paired by use tracks to the west for Chiltern). This means stations for slow trains can just have platforms built either side, and where all four tracks need platforms then only the slower tracks need to dog-leg out. That takes up less space than curving the fast lines, as the curves can be tighter for a lower speed.

Fast Thameslink trains *do* cross from fast to slow and vice-versa on the Midland, though. They usually run fast line from West Hampstead (ish) to Radlett or Harpenden, overtaking a slower one in the process. Yes, it makes timetabling more intricate, as it needs a gap in the traffic on the line that it's crossing, but the trade-off for that is that it makes turnbacks at St Albans, Luton and Bedford much easier without having conflicting moves or expensive flyovers, as well as making it easier to access freight facilities which are all on the slow line side.

The actual reason, though, is really no more than that's how the Midland Railway built it, with a focus on keeping freight traffic separate to passenger, whereas the GN evolved with more of a focus on suburban passenger traffic.

All true, but I'd also point out that the need to cross over a track in the opposite direction in pairing by use makes it more likely that delays in one direction will affect trains in the other direction.

 

[Cricketer8for9]

Paired by use can be problematic for engineering work, especially if there are platforms only on the slow lines. Also a flyover is needed relatively near the terminus to prevent trains having to cross the whole throat.

 

[Horizon22]

It's much more practical during engineering works (access wise and to keep 2 lines running) and disruption as well. Also there's a lot of places where PBD doesn't have platforms on the fast lines where generally (not always, I know plenty of examples) PBU does.

 

[edwin_m]

It's much more practical during engineering works (access wise and to keep 2 lines running) and disruption as well. Also there's a lot of places where PBD doesn't have platforms on the fast lines where generally (not always, I know plenty of examples) PBU does.

Is there actually much difference? With PBU you can close two tracks on one side and there is a "ten-foot"* between those and the other two. With PBD there are usually ten-foots (feet?) either side of the middle pair, so it's possible to close those, though I agree access is more difficult across one of the open outer tracks. This also means that a PBD formation is a bit wider due to the extra ten-foot.

*Known as such, not necessary ten feet but wider than the usual spacing of a double track (which is known as a six-foot). It may be wide enough to work safely on the track one side of it while the other one remains open to traffic.

 

[Pigeon]

The WCML is paired by use from Euston to Rugby (the Slows forming the Northampton loop).

By which it has always seemed to me to be crippling itself, with the cabbaging of the ability for trains in the same direction to overtake one another and all the fiddling about at Hanslope to get ready for the Roade divergence. Better to do it by direction, and have a diveunder for the down slow at Roade.

There are similar constraints on most other routes, too, where they were originally built as a two-track railway and then quadrupled by adding two more tracks along one side. Often these are harder to unscramble than the WCML because of the later tracks following slightly different levels/alignments, or even an entirely separate route in places (eg. Midland). When the slows were expected to be solid with nose-to-tail coal trains, it made more sense like that, but it's less useful for a more varied mixture.

 

[childwallblues]

By which it has always seemed to me to be crippling itself, with the cabbaging of the ability for trains in the same direction to overtake one another and all the fiddling about at Hanslope to get ready for the Roade divergence. Better to do it by direction, and have a diveunder for the down slow at Roade.

There are similar constraints on most other routes, too, where they were originally built as a two-track railway and then quadrupled by adding two more tracks along one side. Often these are harder to unscramble than the WCML because of the later tracks following slightly different levels/alignments, or even an entirely separate route in places (eg. Midland). When the slows were expected to be solid with nose-to-tail coal trains, it made more sense like that, but it's less useful for a more varied mixture.

I would imagine that most trains to/from the Northampton loop would have been sorted by the Milton Keynes stop not Hanslope unless their is engineering works.

 

[Cricketer8for9]

Is there actually much difference? With PBU you can close two tracks on one side and there is a "ten-foot"* between those and the other two. With PBD there are usually ten-foots (feet?) either side of the middle pair, so it's possible to close those, though I agree access is more difficult across one of the open outer tracks. This also means that a PBD formation is a bit wider due to the extra ten-foot.

*Known as such, not necessary ten feet but wider than the usual spacing of a double track (which is known as a six-foot). It may be wide enough to work safely on the track one side of it while the other one remains open to traffic.

Yes, but if they are paired by direction closing the two down tracks means that only up trains can run.

 

[TheDavibob]

By contrast the Tfl line from Finchley Road to Wembley Park runs fast (Met), slow (Jubilee) * 2, fast (Met). However between Baron's Court and Acton Town the line runs slow (District), fast (Pic) * 2, slow (District).

That bit of the Met, including Chiltern, demonstrates *both* approaches and runs V(ery)F-VF-F-S-S-F. The primary reason for paired by direction is to give excellent interchange between the two underground lines, whereas Chiltern running "by use" keeps it out of the way. Horses for courses.

 

[The Planner]

By which it has always seemed to me to be crippling itself, with the cabbaging of the ability for trains in the same direction to overtake one another and all the fiddling about at Hanslope to get ready for the Roade divergence. Better to do it by direction, and have a diveunder for the down slow at Roade.

There are similar constraints on most other routes, too, where they were originally built as a two-track railway and then quadrupled by adding two more tracks along one side. Often these are harder to unscramble than the WCML because of the later tracks following slightly different levels/alignments, or even an entirely separate route in places (eg. Midland). When the slows were expected to be solid with nose-to-tail coal trains, it made more sense like that, but it's less useful for a more varied mixture.

Very little happens at Hanslope. Its all done further south.

 

[swt_passenger]

Paired by use is said to be best at a terminus. Paired by direction is said to be better on a line with multiple flyover or diveunder junctions on both sides. Waterloo to Basingstoke switches from by use to by direction in the Wimbledon area. The pros and cons regarding engineering closures, such as lack of platforms, have been mentioned.

I don’t think it’s a case of paired by use being ‘loved’ or preferred, it is more to do with how some of the four track routes were developed in such a piecemeal fashion, ie they did what appeared to be easiest at the time.

 

[zwk500]

Having experienced the southern commuter belt of both the MML and the ECML, I am massively in favour of quadrupling by direction. However it seems that places me in the minority. The reason I think this is simple: what is fast/slow in one place might not be in another. For example, on the MML, all Thameslink trains have to use the slow tracks, no matter where they are stopping or passing. This means that the capacity for slow trains is greatly reduced while the fast tracks sit fairly empty (EMR only have about 6 trains per hour in each direction out of St. Pancras). This also means that a broken rail (or similar issue) is a nightmare as crossing between the fast and slow tracks is much more inconvenient (which is why the fast Thameslink trains don't do it).

In contrast, on the ECML, fast Thameslink trains can hop between the fast and slow lines with ease, allowing them to switch based on their stopping patterns. For example, the Cambridge-Brighton service joins the ECML on the slow tracks, then after the Digswell viaduct when it expands to four tracks again it stays on the fast lines. Then, at Finsbury Park it crosses over to the semi-fast platform, which then takes it on the slow track towards the canal tunnels. Equally, when there was a broken rail on the up fast a couple of months ago, it was fairly easy for fast trains to cross over to the down fast at the crossing before, and back at the crossing after. Had that been on the southern MML, all hell would break loose.

The other final advantage is based on the fact that realistically, all four-track lines go down to two at some point, whether that's because the line was never quadrupled that far, or because of infrastructural reasons. With quadrupling by direction, this is much more convenient than trying to do this with distance quadrupling, because trains going from one set of tracks to the other don't have to cross another line in the process.

Equally, converging and diverging lines are easier as with directional quadrupling, you can use the following layout:
View attachment 167052
While on distance quadrupling, you have to do this:

(note while the lack of a flyover/under here makes it cheaper, that's only because building this junction graded is effectively impossible)

If you disagree with my assertions, or agree but would like to play devil's advocate, please feel free to do so . I am happy to be wrong as long as I can see why I'm wrong

Paired by use has a significant engineering advantage if you can get all your junctions on one side. The flat junction is less of a constrain if your service pattern allows 'paired' movements. By operating the lines as effectively separate pairs, the problems of crossing back and forth are much lower. It also allows commuter services stopping short to turn back in between the slow lines without the need for any expensive flyovers. For the majority of paired by use lines, the traffic pattern is such that any trains for branches would be using the slows into the terminus.

The main reason for paired by use is more straightforward - it was easier to build 2 tracks alongside and leave the existing railway as was for the 'fast lines' than to reconfigure the railway. The efficiencies of paired by direction require significant distance to become apparent, hence they are more popular on lines with more important longer-distance traffic (ECML, Trent Valley, and South West Main Line). The use of paired by direction on the Windsor lines is, in some respects, quite an odd choice and one where I'd look to convert to paired by use instead.

 

[PeterY]

I don’t think it’s a case of paired by use being ‘loved’ or preferred, it is more to do with how some of the four track routes were developed in such a piecemeal fashion, ie they did what appeared to be easiest at the time.

I have to agree with you swt_passenger . We have to live with how things were built way back when. I think paired by use, or separate, fast and slow each has its advantages and disadvantages.

For me, it is what makes railway infrastructure so interesting. I could name lots of oddities that I find interesting. Sharnbrook slow lines. How the running lines diverge at Princess Risborough. How the Trent Valley lines are signalled for bi-directional running etc

 

[norbitonflyer]

Usually with pairing by direction the fast lines are in the middle, the Met/Jubilee being a rare exception

The Windsor lines between Queenstown Road and Barnes being another, although not rigidly adhered to.

Paired by use can be problematic for engineering work,

On the contrary - it's easier to work on two tracks on one side, keeping the other pair open, rather than closing the outer pair or - particularly, because of access - the middle pair.

 

[edwin_m]

Deleted.

But you can close one up and one down track, both on the same side of the formation, and have two tracks to work on and one in each direction still operating. You would probably only close the two middle tracks when paired by direction.

Incidentally, the relatively recent Trent Valley four-tracking, which is paired by direction in normal operation, has bi-directional signalling on the two middle tracks so the two tracks either side can be closed together.

 

[norbitonflyer]

In contrast, on the ECML, fast Thameslink trains can hop between the fast and slow lines with ease, allowing them to switch based on their stopping patterns. For example, the Cambridge-Brighton service joins the ECML on the slow tracks, then after the Digswell viaduct when it expands to four tracks again it stays on the fast lines.

It is that two-track bottleneck which makes paired by direction the only practical option on the ECML. PBU would require crossovers (on the flat or by flyover) at each end.

If a high speed Welwyn bypass line were ever built, to carry fast trains round the bottleneck ( following the A1(M), or maybe a tunnel under the Mimram valley - modern trains should be able to cope with the gradients needed) a change to pairing by use would be needed to allow the existing station at Welwyn North to still be served in both directions.

You would probably only close the two middle tracks when paired by direction.

How would the workforce access the middle tracks if there were trains running on the tracks either side of them?

 

[edwin_m]

How would the workforce access the middle tracks if there were trains running on the tracks either side of them?

That's the issue with closing the middle tracks, but potentially can be addressed by having a line block on an outer track when access is needed - obviously only possible if the timetable allows for the necessary blocks. Or there might be a station where staff can access via the platform.

 

[Merle Haggard]

Very little happens at Hanslope. Its all done further south.

Interesting that when the high speed Hanslope Junction was built in ca 1972 it was associated with a promise of trains to Northampton running fast to/from the junction.

Is it ever used now where its high speed is crucial? I don't think so, but I'm not sure.

 

[zwk500]

Interesting that when the high speed Hanslope Junction was built in ca 1972 it was associated with a promise of trains to Northampton running fast to/from the junction.

Is it ever used now where its high speed is crucial? I don't think so, but I'm not sure.

It's speed is useful when the fasts are blocked through MK (planned or in an incident). I think there certainly used to be (may still be) some Northampton fasts in the peak that used Hanslope.

 

[Nottingham59]

It seems to me that, in terms of path capacity, the most efficient layout for a four-track railway is with the slow lines in the middle and fast lines to the outside. Such a line could theoretically handle 16tph on the fasts and 16tph or more on the slows.

• On a line with tidal flows, which is all lines out of London, slow trains can terminate and reverse without crossing fast paths
• Fast trains can switch to the slow lines and vice versa without conflicts
• Minor stations only need one platform island, serving both slow lines. (Though platforms on the fasts at some minor stations would add flexibility in times of disruption.)

Other thoughts:

• Station throats at termini would need grade separation to allow 32tph to terminate and reverse off four lines, with enough platforms to handle that level of traffic. Instead, it would be more efficient for a four-track line to go straight though the capital to align with a similar four-track railway on the other side.
• Major stations ideally have six platforms, so that slow trains can terminate in the middle; through fast trains can use the outer platforms; and fast trains switching to the slows at that point can call and wait for a slot on the slows without holding up fast traffic behind.
• Making the two centre lines of four bi-directional would allow the lines to convert to twin-track at night, with engineering access to both lines on one or other side of the alignment. Like is done on the Trent Valley.

Where this arrangement would work:

• I don't see any existing lines which could be converted to this arrangement. TRU chose not to use the layout for its four-tracking from Huddersfield to Raventhorpe.
• This was the philosophy behind the design of the Great Central, with a single island platform at minor stations and space to add fast lines on the outside of the alignment, but that never got the traffic needed to make it work.
• There may be opportunities to use such a layout when quadding lines to increase capacity between Liverpool and York, or between Birmingham and Leeds - when/if NPR and HS2 East get resurrected.
• It could have worked on HS2. Within twenty years, HS2 will be full and another line will be needed north out of London. Provision in the HS2 design for four tracks would have been by far the cheapest way to deliver this.
• In the meantime, a four-track alignment between Ruislip and Wendover could have been built at very little extra cost. At the London end, the fasts could have gone into a tunnel straight to Euston, and the slows use the New North Mainline to terminate at Old Oak Common. At the northern end, the slows would go through Aylesbury to Calvert and then join the Chiltern line at Bicester. Which would have brought fast access to London from Oxford, Banbury, Warwick, etc. in addition to the 18tph on HS2 south of Birmingham Interchange.

 

[Bletchleyite]

Interesting that when the high speed Hanslope Junction was built in ca 1972 it was associated with a promise of trains to Northampton running fast to/from the junction.

MKC Platform 5 would now be used for that, it's connected to the down fast and the down slow. Was Hanslope used before the MKC layout changes of about 10 years ago?

It appears the hourly peak extra fast Northamptons do this, e.g. this one:

Realtime Trains | 1W57 1720 London Euston to Northampton | 10/10/2024

Real-time train running information for 1W57 1720 departure from London Euston to Northampton on 10/10/2024. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

There is one southbound equivalent per day (that I can see), this does still appear to go slow to fast at Hanslope (hence doesn't call at Bletchley or Leighton):

Realtime Trains | 1W02 0539 Coventry to London Euston | 10/10/2024

Real-time train running information for 1W02 0539 departure from Coventry to London Euston on 10/10/2024. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

The other time I know of this being used in this way is when Avanti run via Northampton due to engineering works - again this would be northbound using MKC P5 but southbound moving over at Hanslope. The weekday 2310 New St-Euston (originates from Glasgow I think) is booked via Northampton and does this too.

As for pairing by direction, the advantage of it is that local stations only need platforms on the outside, and that interchange stations can be done as two islands meaning you always have easy cross platform fast to slow interchange (this would be very advantageous somewhere like the south WCML). I guess the downside is that you can't work on one pair separately from the other unless you have bidirectional running and can temporarily change the two ups into an up and a down (or vice versa).

 

[zwk500]

It seems to me that, in terms of path capacity, the most efficient layout for a four-track railway is with the slow lines in the middle and fast lines to the outside. Such a line could theoretically handle 16tph on the fasts and 16tph or more on the slows.

• On a line with tidal flows, which is all lines out of London, slow trains can terminate and reverse without crossing fast paths
• Fast trains can switch to the slow lines and vice versa without conflicts
• Minor stations only need one platform island, serving both slow lines. (Though platforms on the fasts at some minor stations would add flexibility in times of disruption.)

Other thoughts:

• Station throats at termini would need grade separation to allow 32tph to terminate and reverse off four lines, with enough platforms to handle that level of traffic. Instead, it would be more efficient for a four-track line to go straight though the capital to align with a similar four-track railway on the other side.
• Major stations ideally have six platforms, so that slow trains can terminate in the middle; through fast trains can use the outer platforms; and fast trains switching to the slows at that point can call and wait for a slot on the slows without holding up fast traffic behind.
• Making the two centre lines of four bi-directional would allow the lines to convert to twin-track at night, with engineering access to both lines on one or other side of the alignment. Like is done on the Trent Valley.

Where this arrangement would work:

• I don't see any existing lines which could be converted to this arrangement. TRU chose not to use the layout for its four-tracking from Huddersfield to Raventhorpe.
• This was the philosophy behind the design of the Great Central, with a single island platform at minor stations and space to add fast lines on the outside of the alignment, but that never got the traffic needed to make it work.
• There may be opportunities to use such a layout when quadding lines to increase capacity between Liverpool and York, or between Birmingham and Leeds - when/if NPR and HS2 East get resurrected.
• It could have worked on HS2. Within twenty years, HS2 will be full and another line will be needed north out of London. Provision in the HS2 design for four tracks would have been by far the cheapest way to deliver this.
• In the meantime, a four-track alignment between Ruislip and Wendover could have been built at very little extra cost. At the London end, the fasts could have gone into a tunnel straight to Euston, and the slows use the New North Mainline to terminate at Old Oak Common. At the northern end, the slows would go through Aylesbury to Calvert and then join the Chiltern line at Bicester. Which would have brought fast access to London from Oxford, Banbury, Warwick, etc. in addition to the 18tph on HS2 south of Birmingham Interchange.

This precise arrangement is used in the Republic of Ireland, for their short section of 4-track between Park West & Cherry Orchard and Hazelhatch and Celbridge.

One problem with having the fast lines on the outside is that the 'splay' to accommodate the central island platform must be more gentle the higher speed you go. One reason why paired by use has been quite popular is that it allows 4-platform stations to take up a limited footprint as the fast lines continue through and the slow lines can make the compromise of a tighter curve between the island and the outer platform.

 

[Nottingham59]

This precise arrangement is used in the Republic of Ireland, for their short section of 4-track between Park West & Cherry Orchard and Hazelhatch and Celbridge.

I see it's also used on the Underground, between Finchley Road and Harrow-on-the-Hill.

One problem with having the fast lines on the outside is that the 'splay' to accommodate the central island platform must be more gentle the higher speed you go. One reason why paired by use has been quite popular is that it allows 4-platform stations to take up a limited footprint as the fast lines continue through and the slow lines can make the compromise of a tighter curve between the island and the outer platform.

Yes, that's a good point. I hadn't considered the splay aspects.

 

[Bletchleyite]

Yes, that's a good point. I hadn't considered the splay aspects.

If you have the fasts in the middle you can of course just splay the slow lines and keep the fasts perfectly straight. And no splay is needed at stations which only have slow line platforms (though if you do that you can't serve those stations if the slows are not in use).

 

[mr_jrt]

It seems to me that, in terms of path capacity, the most efficient layout for a four-track railway is with the slow lines in the middle and fast lines to the outside. Such a line could theoretically handle 16tph on the fasts and 16tph or more on the slows.

My personal preference is also the GC-style, but your list above omits the reason it is never seen anywhere else - the land required for the fast lines to bulge out without massive speed reductions.

As a thought experiment, I once considered conversion of the southern WCML. Given the lack of modern junctions, I think you could convert it, if you could acquire the land required so as not to reduce line speeds. More practicable I suspect would be to convert it the other way, with the fasts in the middle.

All stations have suitable platforms, and really you just need new grade separation at the junctions at Roade (& probably Bletchley), and obviously at turning points like Euston, and Tring. Obviously there would also be horrific signalling work (ECTS might make that feasible), and lots of crossovers etc that would need changing, but not entirely impossible...just extremely unlikely to ever wash its face economically.

Suffice to say, interesting thought experiment, but nothing more.

EDIT: I see the splay has come up - took too long typing!
EDIT 2: Just occurred to me, in my WCML example for FSSF, you would only actually need new land to realign the current up slow (proposed up fast), as the existing down fast is already laid out suitably. Doesn't massively change things, but worth mentioning nevertheless.

 

[MarlowDonkey]

• This was the philosophy behind the design of the Great Central, with a single island platform at minor stations and space to add fast lines on the outside of the alignment, but that never got the traffic needed to make it work.

On the GC & GW joint, they used the opposite philosophy. Most of the stations were built with platforms on a loop, so the through lines were in the middle. At many of the stations, Gerrards Cross being an example, the London direction platform has now been extended outwards. You can see where the loops used to be at High Wycombe and Beaconsfield as the through line has been relaid to be the former loop.

 

[Cricketer8for9]

But you can close one up and one down track, both on the same side of the formation, and have two tracks to work on and one in each direction still operating. You would probably only close the two middle tracks when paired by direction.

Incidentally, the relatively recent Trent Valley four-tracking, which is paired by direction in normal operation, has bi-directional signalling on the two middle tracks so the two tracks either side can be closed together.

But one up and one down can’t be on the same side of the formation with paired by direction.

 

[furnessvale]

That bit of the Met, including Chiltern, demonstrates *both* approaches and runs V(ery)F-VF-F-S-S-F. The primary reason for paired by direction is to give excellent interchange between the two underground lines, whereas Chiltern running "by use" keeps it out of the way. Horses for courses.

As you say, horses for courses. As an example, leaving Man Picc, you have Oxo Rd entering from the right, Guide Bridge departing from the left, Heaton Norris joining from the left, Chester departing from the right, Hazel Grove departing from the left and Cheadle Hulme departing from the left, all within 7 miles or so.

In the absence of flyovers, which pairing is better, use or direction, and if paired by use, which side do you put the slow lines on?