HS2 – where now?

Recent government announcements effectively put HS2 back to being Phase 1 only, that is, new high speed railway from London (Euston) to the West Midlands, running to a new junction with the West Coast Main Line at Handsacre.

Reversion to a ‘Phase 1 only’ version of HS2 brings a number of operational problems, with expensive solutions, that the Government decision can hardly have been able to work through. Given that, it is borderline irrational to rule out Phase 2A or something very like it. Overall, the lack of understanding within the industry as a whole as to how HS2 would work as a railway still surprises me, but I think the responsibility for that lies firmly with the DfT, who have always avoided anything that might look like a commitment to detail of a particular service option. Whilst it seemed from 2010 onwards that every conference had a presentation about the engineering of HS2, it took me seven years to gain permission to publish a single article on its potential operation and services!

“Phase 1 has not been a ‘thing’ since Phase 2A was defined, so what do we need to resurrect from HS2’s early analysis, and what does it tell us?”

That may be understandable, especially in the face of opposition groups trying to pick holes in the proposed services on the basis that this, that or the other station would lose out, the most ridiculous such attempt being based on an idea that Wolverton would lose services when there is no logical relationship whatsoever between Wolverton and HS2. But the result is that discussion of HS2 can still descend into complete confusion, and recent Transport Select Committee hearings sadly have produced examples of this.

Phase 1 has not been a ‘thing’ since Phase 2A was defined, so what do we need to resurrect from HS2’s early analysis, and what does it tell us?

Phase 1 timetabling

Timetable planning needs planning rules, Sectional Running Times, a Train Service Specification, and a dollop of luck! Some things work, some don’t; the skill is recognising the difference and working with it.

Feasibility timetabling of HS2 Phase 1 services was undertaken according to Train Service Specifications as published in the various Business Case documents, as well as some variants tested and rejected before incorporation into a Business Case. Of ten trains per hour (tph) on HS2 from Euston, seven would use the Handsacre junction – three Manchester, one Glasgow, one Preston, two Liverpool. Planning rules and Sectional Running Times were determined by simulation.

HS2 Phase 1 service pattern diagram

Iteration with HS2 engineers and the DfT led to changes being made progressively both to the specification and to the infrastructure. As a personal observation, this was the best example of such a process in all my experience, and my summary of the operations planning aspects was published in Modern Railways in April 2021.

Timetable planning is a continuous process of balancing requirements and opportunities. A specification must be defined for purposes of demand modelling, but changes will be made on the basis of findings from both that modelling, and operations planning to test the feasibility of the services. This process supports the Business Case by defining not just demand but also operating costs such as crew numbers and fleet size, but will continue almost to the implementation date.

“Timetable planning is a continuous process of balancing requirements and opportunities.”

For these reasons, it is not valid to make statements such as that the HS2 service ‘will’ or ‘won’t’ do this, that or the other, serve or not serve station x or station y. As examples:

  • It has been said that a number of stations would lose London services, such as Warrington and Lancaster. In the case of Warrington, an HS2 service has always been part of the specification and always found to be feasible; in the case of Lancaster an HS2 service was found to be feasible given Phase 2A running times, and incorporated into the specification;
  • In the demand modelling, HS2 Manchester trains were assumed not to serve Crewe. However, with the Fast lines connection, this is physically possible, and whether it is done or not is a matter for demand and for operating feasibility with respect to issues such as junction conflicts and terminal turnrounds.

Whether other locations such as Oxenholme and Penrith are served by HS2 trains is a matter for demand and operating feasibility – these are choices, not laws, and the timetable for ten or more years hence will present these trade-offs whether HS2 is built or not. It is the role of the West Coast Partnership to devise potential train service plans for HS2 and conventional services, and they will be exploring all these trade-offs.

How about the rest of the railway?

In 2013/4, I was tasked with undertaking a timetabling exercise to demonstrate integration of HS2 Phase 1 services with a residual WCML conventional service in which the bulk of long distance demand was catered for by HS2 services. The conventional specification was developed by an industry working group of train operators and the DfT, and at Handsacre featured, in addition to HS2 services:

  • 1 tph Euston – Chester as now
  • 1 tph Euston – Manchester, serving intermediate flows such as Milton Keynes – Manchester and Euston – Stoke-on-Trent;
  • A second Euston – Manchester service as above in peak hours only;
  • 1 tph Euston – Crewe serving Trent Valley stations such as Tamworth and Lichfield;
  • Four freight paths per hour in each direction between Crewe and WCML locations South of Handsacre (not necessarily London area but also East Coast via Nuneaton, and Daventry/Northampton);
  • Two freight paths per hour between Crewe and the Wolverhampton line at Stafford.

It proved just possible to timetable this specification with respect to a study area from Euston to Crewe, albeit with some departures from the specification at peak times. Reliability of operation was not tested, but would have been questionable. A specification devised today would likely be more ambitious especially in respect of freight. Additional services such as a second Trent Valley stopping service were not specified, but would not have been feasible.

The key constraints were found to be Colwich Junction, the double-track section through Shugborough tunnel South of Stafford, flat junctions at Stafford, and Crewe station. A reasonable timetable on the conventional WCML Fast lines from Euston could feature 14 trains per hour, but this capacity could not be used to the full unless all these capacity constraints are evaded so as to provide trains leaving Euston with somewhere to run to. Shugborough in particular stands out, and quadrupling it seems an obvious solution. But this wasn’t done under the West Coast Main Line Upgrade, and for very good reasons.

“Capacity could not be used to the full unless all these capacity constraints are evaded.”

In the WCML Upgrade project completed in 2010, most of the WCML from London to Crewe was indeed provided with four tracks. Two possible configurations of a four-track railway, termed ‘paired by use’ (Up-Down-Up-Down) and ‘paired by direction’ (Up-Up-Down-Down) are possible. Each has its advantages and disadvantages, but the choice is largely determined by historical ease of construction, and both are seen on the WCML.

Transitions between the two configurations either incur conflicts between trains in opposing directions (Manchester Slade Lane Junction) or flyovers/dive-unders (Wimbledon, Camden). In fact part of the complexity of Crewe derives from the transition on the flat between ‘paired by use’ South of the station and ‘paired by direction’ North of the station.

And if quadrupled from Colwich to Stafford, a transition between ‘paired by direction’ and ‘paired by use’ would arise, implying either conflicts between train paths limiting capacity, or a further grade-separation somewhere.

The limitations on the train service imposed by feeding HS2 trains straight into these capacity constraints was one of the reasons why Phase 2A was defined.

If you’re enjoying this… why not watch:

HS2 Phase 2A

What became Phase 2A was the West Midlands to Crewe section of Phase 2, separated out with its own Hybrid Bill, and intended to be opened simultaneously with Phase 1.

Under the Phase 1 Hybrid Bill, the connection at Handsacre was to be with the WCML Fast lines, to minimise interference with other traffic and obtain the best possible running times. As the configuration at this point is ‘paired by direction’, the Fast lines are the two middle tracks at this point, this connection implied diverting the Up Fast and Up Slow lines onto a new alignment so as to create space for the flyover from HS2 to land between the Up and Down Fast lines.

When Phase 2A of HS2 was defined, it was seen that six of the seven HS2 services would run direct to Crewe. The proposal for Handsacre junction was thus simplified to make connections with the Slow lines. As these are the two outside lines on a section ‘paired by direction’, realigning the Up Fast and Up Slow was avoided, saving cost and disruption.

Phase 2A would then connect into the WCML South of Crewe. A number of configurations for the connection were tested, but the final plan was for the connection to be with the Fast lines, with crossover connections to the Slow lines to allow trains to access the station platforms. The earlier ‘Slow line’ connection derived from an over-literal interpretation by a design consultant of a requirement for trains to be ‘able to access’ the Slow lines.

Amongst other advantages, the Fast line connection translates directly into the ‘paired by direction’ configuration of the WCML North of Crewe, resolving a long-standing constraint.

Phase 2A timetabling

As for Phase 1, timetabling for Phase 2A was undertaken on the basis of specifications set by the DfT. Compared with the Phase 1 specification, the level of service remained at 10 tph from Euston, but apart from reduced journey times exploiting the 2A infrastructure, changes were:

  • Preston service extended to Lancaster. Note, the ability to do this derived specifically from the reduced running times given Phase 2A, and whether it could be done given reversion to Phase 1 is doubtful;
  • This Lancaster service to run combined with a Liverpool train from Euston and divide at Crewe;
  • The path freed by running the Lancaster and Liverpool trains combined was redeployed for a service calling at Stafford, Stoke on Trent and terminating at Macclesfield;
  • Availability of the Macclesfield service to serve Stafford allowed both Liverpool trains per hour to run via Phase 2A and Crewe, providing an even half-hourly service at Liverpool rather than one train per hour still having to run via Stafford.

It was a finding that many things in the HS2 plan for Phase 2A worked very well. For instance, terminal turnrounds all fell into the category of ‘adequate but not excessive’. Another very useful feature was that the Manchester services fell ‘in parallel’ at Crewe North Junction, that is, as a Down Manchester service crossed the junction, an Up Manchester service did so on the opposite line, using a path that would be denied to a train on any other route. Thus, although the number of trains using the junction increased in absolute terms, this was offset by improved efficiency of use of the paths across the junction.

Alternatives to Phase 2A

The difficulty of quadrupling through Shugborough tunnel has been discussed earlier, in the form of choosing a configuration for the quadrupled route without creating conflicts at the transition to existing four-track section. To make any use of such a quadrupling, grade-separation would also be required at Colwich and Stafford. In effect, the WCML would be converted wholesale to ‘paired by direction’ from Colwich through Crewe, essentially a complete redesign, reconfiguration and rebuild of the whole railway, on a route for which there is no reasonable diversion available for the Phase 1 level of service.

In planning for the WCML Upgrade completed in 2010, a ‘Stafford Bypass’ was devised by Arup, and described to me by my good colleague Phil Hall. This bypass would have commenced immediately north of Rugeley Trent Valley station, where a length of straight track would have allowed slewing of the Up Slow and Up Fast to create an island between the Fast Lines. The bypass started in this island, grade separated over the slewed Up lines. It then passed north of Great Haywood, exactly where Phase 2a was proposed, then over the A51, over the Colwich to Hixon line, and over the Trent.  It then went through Ingestre Golf course, and just south of the village of Hopton, just like 2a. It then deviated from 2a, and headed towards the north of Norton Bridge, crossing the A34 and M6, before connecting to the WCML Fast lines near Coldmeece. Only small journey time savings were expected, given conventional speed connections at each end.

Land-take and compulsory purchase powers would be required for any of the above alternatives.

Economic Case for Phase 2A

I understand from Freedom of Information that the latest Benefit:Cost Ratio (BCR) for Phase 2A as an increment to Phase 1 is 1.3.

Whilst the reason for building new infrastructure of any sort is capacity, the benefits are maximised when the new infrastructure allows for high speed, and reduced running times will provide benefits compared with Phase 1.

The BCR will improve when it is considered that extra costs at Handsacre to recreate the Fast Line connection will be avoided, and that rather than being a cost to Phase 2A, much of the work at Crewe is effectively in the ‘Do Minimum’ case. Moreover, reliability of operation will radically improve which will bring further measurable benefits. Additional train mileage will be incurred for the more demanding specification, but justified by demand, whilst the fleet needed for the extra services will largely be provided by greater efficiency deriving from reduced round-trip times.

Splitting and joining of trains

Economising on trunk route capacity by running trains for two destinations combined, and splitting at a junction station, is a long-standing option in rail operations. Indeed the HS2 service specification has already featured limited resort to this practice. However, it brings performance risks, both from the possibility of a technical failure particularly when rejoining on the return journey, and operationally as one portion arriving late to rejoin delays both portions on departure.

The practice also imposes a constraint on timetable planning, in that a relationship is forced between two trains that may then limit other options. For instance, if a Manchester and Liverpool train were to run combined to Crewe, their timings on departure from Crewe would be very close to each other, with the possibility that a timing selected to suit the congested Manchester approaches might lead to a sub-optimal solution on the congested Liverpool approaches, and vice-versa.

Additional time must also be allowed for the tasks of splitting and joining, which adds to journey times, especially where an additional stop has to be made for the purpose. In the case of HS2 Phase 1, a timetable exercise indicates that when achieving a 100-minute journey time to Manchester by calling only at Stockport with one of the three trains per hour, and allowing the other two to call at either Wilmslow or Stoke on Trent, turnrounds at Manchester Piccadilly are between 29 and 31 minutes. Time lost en route by making an additional stop at Crewe to split would add between seven and ten minutes to each leg of the journey, leaving turnrounds at worst of potentially as little nine minutes at Manchester, which is completely unrealistic. The need then would be to introduce another train into the circuit so that longer turnrounds could be taken, adding to the fleet size and potentially increasing platform occupation at this already critical location.

The risk is even greater at Liverpool, where expected running times already lead to turnrounds of only just over 20 minutes. This is not to say that a planning mitigation and compromise cannot be found, simply that such a mitigation and compromise must be found before committing to and relying on this method of operation.

To tilt, or not to tilt?

One thing the TSC managed to get its proverbials in a twist about was the issue of whether or not HS2 trains should tilt as Pendolinos do. In fairness the answer “no they shouldn’t” was pretty clear when the ultimate scope of HS2 was to feature new railway almost as far as Preston, leaving relatively little running on the twisty bits of the WCML thence to Scotland. But if Handsacre, for good or ill (mostly ill of course) is to be the limit of new infrastructure, the amount of WCML running increases to about two-thirds of a Euston to Glasgow journey.

The TSC was I think seriously misled by the MD of Angel Trains, who told them that ‘like for like’ the benefit of tilt on that journey was about 20 to 24 minutes. Now I haven’t seen the latest modelling, but that strikes me as incredible. The likelihood as I see it is that the figure quoted is the difference between a Pendolino running time after the WCML Upgrade, compared with the previous ‘loco + coaches’ timings. That is, after all, what Angel Trains would know.

“One thing the TSC managed to get its proverbials in a twist about was the issue of whether or not HS2 trains should tilt as Pendolinos do.”

But that estimate involves not only tilt, but also a top speed of 110mph even on the straight bits. And although the DfT declared that in order not to spend a penny on the conventional railway, not even to change a speed board from ‘110’ to ‘125’, there are significant sections of the WCML North of Handsacre where an increase in permitted speed could be achieved without tilt.

Much in terms of journey time depends on the assumed stopping pattern, and my estimate of a Euston – HS2 – Handsacre – Glasgow journey time with an increased permitted speed, and stops at Preston and Carlisle (remember the Lancaster train is there to pick up some of the intermediates) is four hours seven minutes, from which I estimate tilt would deduct a further five minutes. Given that adding the tilt equipment would, apart from the modification to the trains, add weight to be carted around day in day out wherever the train goes, possibly increasing the axle-weight to a point where speed on the new infrastructure has to be limited, is that worth it?

Beyond Phase 2A

A key capacity constraint not addressed in any way by the document ‘Network North’ is platform capacity at Manchester Piccadilly station, where already the station operation depends on putting two or even three trains into the same platform. This is operationally fragile, and confusing for passengers. Moreover, it becomes impossible to lengthen trains, as the total train length risks exceeding the platform length clear of the station throat.

London trains take up two platforms at Piccadilly, every hour of the operating day. Under Phase 2B, London trains would have been removed from Piccadilly station, freeing two platforms for other uses.

“If we get Crewe right, there is scope for economy in some aspects proposed by HS2 Ltd, which will improve the BCR.”

From Freedom of Information, I understand that the BCR for Phase 2B as an increment to Phase 2A is 0.6. Clearly this does not support the project. However, in my opinion, if we get Crewe right, there is scope for economy in some aspects proposed by HS2 Ltd, which will improve the BCR.

Moreover, the document ‘Network North’ reserves £12 billion pounds for Northern Powerhouse Rail (NPR) to build effectively the same infrastructure as HS2 would have done from Hoo Green through the Manchester Airport station to Central Manchester. If this infrastructure is in place and the cost not thus attributable to HS2, adding the Crewe to Hoo Green section of HS2 would have a BCR that I estimate at a very compelling 4. This beyond doubt justifies filling the Crewe to Hoo Green gap especially as doing so then frees platform capacity at Manchester Piccadilly.

If you’re enjoying this… why not watch:


With the HS2 Phase 1 service of seven trains per hour, and the conventional specification tested in 2014, the WCML between Handsacre and Crewe will be ‘full’, in that nothing can be added unless something else is removed. Growth beyond that very unambitious specification will be impracticable, and even at that level of service acceptable reliability of operation is doubtful.

HS2 Phase 2A would bypass the key capacity constraints at Colwich Junction, the double track section through Shugborough tunnel, flat junctions South of Stafford. At Crewe station it would allow a radical simplification of operations improving both capacity and reliability.

Any conventional approaches to easing these capacity constraints will be disruptive over and above the pure engineering cost, and require land purchase powers.

An operation based on extensive splitting and joining of trains must be tested rigorously for both planning feasibility and reliability of operation before being adopted as a palliative for capacity constraints.

If tested against a realistic ‘do minimum’ scenario in which Crewe remodelling of some form is assumed, and when considering the cost avoided at Handsacre, the economic case for Phase 2A can be expected to be strong.

Whether this economic case is or is not stronger than the case for improving capacity between Handsacre and Crewe by remodelling at Colwich, Shugborough and Stafford, or by reversion to the ‘Stafford Bypass’ scheme, has not been tested.

Safeguarding of the Phase 2A route, much of which is common with the Stafford Bypass, must not be lifted, nor land disposed of, until that analysis has been undertaken and unless the alternatives to Phase 2A are found to be more favourable to provide for reliability of operation and fill gaps in the conventional specification in the short term, and for further growth in the longer term.

If the NPR project justifies expenditure on a new Western approach to Manchester, it is logical then to plan to enable HS2 trains to use this infrastructure, to free capacity at Piccadilly station, improving the case for, and maximising the benefits of, both projects.