Cambridge Approaches

Introduction

The article below has been written by an anonymous Cambridgeshire contributor who has a keen passion for, and knowledge of, railways. The previous article on this subject showed that the general capacity of the existing twin track between Cambridge Station and Milton Junction was sufficient to take the existing traffic, planned growth and EWR trains without the need for 4-tracking.

In discussions and correspondence with EWR Co, it was clear that they have carried out their assessment based on the current timetable. They claim that this showed that 4-tracking of this section of the track would be required. Our previous article suggested that this was unnecessary as the current timetable would change over the next 10 years before EWR became operational.

However, to address EWR Co’s specific concerns, we have looked at the issues that they raise with the timetable and show that, with a minor revision, even on this basis there is no need to 4-track this track.

Summary

EWR Co’s argued need for 4-tracking in the north of Cambridge in a ‘northern approach’ scenario – which underpins their decision to discount that option in the latest consultation – is based on concerns about interfacing with current operations. EWR Co have not published their full analysis, but state that they cannot fit EWR trains at regular 15 minute intervals alongside the existing timetable. While that statement appears to be technically true (in the most tenuous sense), the analysis presents here suggests that without context, it is likely highly misleading – and 4-tracking is certainly not the only solution. The issue is not too many trains, but a few quirky ‘symmetry breaking’ inefficiencies in the current timetable (particularly the fast/slow Liverpool Street trains). The immediate EWR issue can be solved by slipping a single hourly train by a single minute, and even if we are to future-proof to include the reasonable-worst-case scenario (including a variety of other future trains beyond EWR), only very minor tweaks to the existing are needed to make it work. Once Cambridge to Liverpool Street services are expanded to four an hour (expected in a similar timeframe to EWR’s construction), the asymmetry issues will have resolved themselves anyway – so it is illogical to let such quirks control EWR’s design decisions.

1.      Introduction

(I suggest you grab a cuppa and get comfortable before you continue, as this is quite a long and technical post)

I should start by saying that I’m not a campaigner and am not necessarily pushing for any particular outcome. Having seen the significant campaigning efforts of Cambridge Approaches, CBRR and others, however, I thought it would be interesting to delve into the ‘north vs south’ arguments from an objective viewpoint, to try and determine whether or not the proposed northern approach is actually a viable alternative. A few months of late-night analysis later, and yes, I do think there are – potentially – some aspects which may have been overlooked. I am not an industry expert and am of course willing to be stand corrected if I have missed something important, but am keen to open the discussion so that we maximise our chances of building the best railway possible. Although EWR Co have published many pages of technical documents describing the consequences of 4-tracking, they have unfortunately still not provided a robust technical justification for why they think it is necessary – instead simply claiming that they have “looked at it and concluded that it’s the only solution”. This makes it very difficult to peer review their work and assess the validity of their argument, so some guesswork is required in attempting to re-trace their steps. I’ve tried to present everything as clearly and transparently as I can in this document, so people can see the arguments for themselves.

If, as EWR Co claim, the only way to add four hourly EWR trains through Cambridge North is to add a second pair of tracks through the north of Cambridge, then the associated destruction of nearby properties and other infrastructure costs are very significant (as the infamous “appendix F” goes to great lengths to illustrate). These consequences feature front-and-centre in EWR Co’s justification for discounting a northern approach, and I suspect finding a way to avoid 4-tracking is at the core of the northern approach’s viability.

If my analysis has missed some vital other reason why 4-tracking really is the only solution, then the proposed southern approach certainly has its relative merits. However, there is still a possibility that as even more Cambridge-London trains are added, sending EWR trains into Cambridge from the south might mean it isn’t long before the track from Shelford to Harston needs 4-tracking, alongside grade separation of the Shelford junction, both of which would also be highly disruptive.

2.      Mapping the new railway

In the vector diagram below, every line corresponds to a single hourly train in each direction at present (except for the orange freight line, which is just indicative of routes).

The next diagram is a best guess at how EWR Co’s proposed ‘southern approach’ might look. This analysis is based on an assumed 4 trains an hour, which would be in line with the busiest UK inter-city services running today. I’ve included the potential Wisbech line (either running from Cambridge or splitting from existing trains at Ely), added second hourly trains to both Norwich and Ipswich, and extended a second hourly London train from Ely to King’s Lynn. This gives us a reasonable upper bound for the next few decades to the north and east. I’ve not added potential extra trains south of Cambridge, such as the mooted services to Maidstone or Haverhill, or additional London services (at least three of which are included in Network Rail’s 2030/40 scenarios). There is still some uncertainty about how EWR would be timetabled: I’ve taken the ‘best case’ scenario where Norwich keeps its direct Stansted service, but otherwise 3 of 4 EWR trains continue to Norwich (1/h) and Ipswich (2/h). EWR, then, serves Cambridge South and Central stations with every train; Cambridge North only sees a single hourly EWR service.

Next, we have a potential ‘northern approach’. The same extra services are assumed as above, but EWR now arrives via a junction at Milton rather than Harston. There’s an optional station at Northstowe (and space still for one at Harston too), and 4-tracking may only need to go as far as Cambridge South. In this scenario, the ‘base case’ is that EWR trains all serve all three Cambridge stations, terminating at Cambridge South. There are options, such as linking with existing London (if electrified), Stansted and/or a reopened Haverhill line, or reversing at Cambridge in order to continue towards Norwich/Ipswich. It is a requirement that EWR’s design does not preclude the latter as an option, though it is probably not the most beneficial solution. Freight mostly avoids Cambridge via a new chord south of Ely. This would, admittedly, be expensive to construct (as it would include a new bridge over the Cam), but could remove freight from the single-track Newmarket route, reducing the need to dual quite as much of it.

3.      Improving existing layouts

It is clear from the northern approach diagram that the section through the north of Cambridge is quite busy – so at first glance the argument for 4-tracking seems reasonable. But can it be avoided? The first step is to see what can be done, with only modest adjustments, to increase efficiency in that section. South of Cambridge (central) station we will have 4 tracks, a grade-separated Milton junction confers no additional capacity constraints, so the solution hinges on just the three miles of twin tracks between Cambridge’s central and north stations.

Two hourly Liverpool Street trains terminate at Cambridge North (except for 2/day to each of Ely and King’s Lynn). When they head back south, they block the northbound track until they reach the switch point south of the river. However, this can easily be avoided by moving the terminus platform from the west to the centre. Although it should not be necessary, the through-platforms could be doubled up: There’s space for the western one within the existing footprint, while a 5^th on the eastern side will probably take a sliver of land from neighbouring properties. In the very worst-case scenario, if no extra land is available, then there is space further north to stagger northbound and southbound platforms. If, as has been suggested, the two Liverpool Street trains will in future need to continue to Ely, then a terminus here is not required, and either 2 or 4 platforms would suffice.

Cambridge Central is a large and complex station, currently having four terminus and four through-platforms (the latter all serving trains in both directions). The terminus platforms are all on the western side, so reversing trains again obstruct oncoming traffic as they switch tracks. According to Network Rail’s 2019 report, a southern approach would see a new island pair of through-platforms to the east, and 3-tracking to the Coldham’s Lane junction, as shown on the right in the figure below.

There are many possible iterations for upgrading the station in a northern approach scenario. The first obvious step is to swap Ipswich/Newmarket trains onto a new eastern platform. Without the need to weave Newmarket-bound trains across from the eastern side of the Shelford junction as they approach from Oxford, it should be possible to separate them entirely from other services via 3^rd track to the east towards the Coldham’s Lane junction. Ipswich trains currently only dwell at Cambridge for about five minutes, so even with two or more such services an hour, a single platform and single track under Coldham’s Lane bridge should be sufficient, assuming most freight is diverted via a new spur near Ely.

It may make sense to tidy the central portion of the station into a pair of northbound and a pair of southbound platforms. In a similar approach to Cambridge North mentioned above, it might also be possible to allocate a terminus-priority platform in the middle (the current platform 7) which could eliminate at-grade crossings on the section we’re interested in. This platform could be extended to the south, to also provide a central terminus for trains from London. If the latter option is chosen, platforms 5 and 6 might even become redundant: paved over, they’d make the ideal location for a second station entrance, providing a much shorter route from cycle parking to the platform bridge, and additional income from renting out retail space.

It is worth noting that Cambridge central station is more constrained to the south than the north: the Hills Road bridge is just 100 metres beyond the end of platform 1 (and only four tracks wide), and the post office building & car park limits space to the southeast. At present, there is space for the existing platform island to be extended further south – though if the new eastern platform island were connected to the south, this option may become less viable. In a northern approach scenario, there is much better balance in the number of trains heading north vs south in and out of Cambridge, which should reduce the pressure on platform space and time from terminating services.

For completeness, we should mention Cambridge South. In the southern approach proposal, it has four platforms – two pairs serving ~14/h in each direction (plus any future London services). In a northern approach scenario it may serve slightly fewer trains, but some might terminate at the station. It should be possible for a single platform to cope with as little as 3 minutes between trains (see Tottenham Hale’s southbound platform 1, for example), but in the worst-case scenario a 5^th platform may be required here, offering a central terminus and two through-platforms in each direction. Although there would be some opposition to encroachment on Hobson’s Park, the additional land and cost would be very modest (just the width of a single-track path, as the current proposed station design already has the requisite platform island). If few EWR trains terminate here (instead continuing south to a combination of Haverhill, London, Stansted or elsewhere), the chance of needing a 5^th platform is reduced, but 4-tracking may instead need to continue to the Shelford junction – as is the case for a southern approach scenario.

With a little effort, then, it should be possible (if needed) to entirely eliminate at-grade crossings on the busy section we’re interested in, between Cambridge and Cambridge North. We therefore have a simple scenario of twin tracks, around a dozen trains per hour, and no clashes or constraints linking the two directions. Fewer freight trains should need to use this route (with as many as possible moved to EWR and an Ely chord), so trains should follow very similar acceleration profiles on this short, slow section of track. Nonetheless, we will still aim to ensure one oversized hourly slot remains for freight just in case (because getting as much freight off of the road and onto rail is very much to be encouraged).

If we look at the existing timetables, we see that the minimum gap between trains leaving Cambridge to the north is 3 minutes (xx:00 and xx:03 Cross-Country and Great Northern respectively). A 3-minute headway is also what’s stated in EWR’s requirements [Appendix B 5.7.2 of the latest EWR Co consultation docs] – so we can safely assume this is our minimum acceptable gap. Indeed, it only takes about 3 minutes between leaving one station and arriving at the other, so even spaced this close together, we should never have two trains between the two stations at the same time.  

I should note that the Cross-Country train does not currently stop at Cambridge North station – so despite many examples to the contrary, such as Tottenham Hale mentioned above, EWR Co may still conclude in the worst-case scenario that Cambridge North requires expansion to twin platforms in each direction. This is an expense not required with a southern approach, but that’s only because in a southern approach few (if any) EWR trains actually serve Cambridge North.

4.      Symmetry and inefficiency – the spectre of 4-tracking

In an idealised world without external constraints, 3-minute headings should allow for a theoretical maximum of 20 ‘slots’ per hour (up to 15 of which can be used if we are to stick within the 0.75 resilience requirement). With this in mind, EWR Co’s argument that even the 6 current + 4 EWR trains cannot be accommodated without quadrupling the tracks has led to much head-scratching. So, what is going on; how is the real world so far from ideal that we can’t even manage what is – without even considering future signalling upgrades – just ~50% theoretical capacity?

Although EWR Co have still not published their actual analysis (only the consequences thereof), I recently managed to speak to EWR Co’s Paul Sparrow via their live chat events, and obtained slightly more insight into their analysis:

“We have used some operations specialists to conduct the assessment using rail operations modelling systems in order to ascertain where we could get our 4 trains per hour onto the existing twin track WAML using our regular 15 min clockface service pattern. These results demonstrated that the white space, the space between trains, does not exist to allow us to get 4 trains an hour every 15 mins.” (Paul Sparrow, 29^th May 2021)

This snippet of information, although still quite cryptic, is perhaps just enough for us to solve the mystery. The issue, I now suspect, is actually quite simple – and it’s all down to symmetry. Despite computational advances, building railway timetables is still ultimately mathematically driven by repeating patterns: it’s what the Swiss call ‘taktfahrplan’, or clock-face scheduling. Most train lines interact with several others along the length of their routes, and weaving all of these interconnected requirements together without clashes is much easier if everyone is dancing to the same beat. Although modern timetables may look quite random, they are essentially still just perturbations on an underlying regular symmetry.

We can illustrate this with a simplified example. Consider a railway line serving four locations: Appletown and Bananaville each need four trains an hour; Carrotbridge and Donutland need two apiece. Treating C&D as a single set, we can schedule these 12 trains at 5-minute intervals, with a pattern repeating every 15 minutes (left clock-face below). However, what if Carrotbridge decide they need their pair of hourly trains spaced 20 minutes apart? We start with Appletown’s regular trains, but once we’ve added Carrotbridge’s symmetry-breaking requirements, there’s no space left for Bananaville to have a train every 15 minutes. We can squeeze in the pair of Donutland trains, but the remaining slots are unhelpfully distributed – poor Bananaville!

This scenario is conceptually very closely related to the issue we have at hand. Carrotbridge in this case is represented by the Greater Anglia trains to Liverpool Street – and Bananaville is Oxford!

The two hourly trains from Cambridge North to Liverpool Street arrive in London at regular half-hour intervals. However, one of them is slightly ‘fast’ and the other ‘slow’: the difference is a modest ~10 minutes (~1h10 vs ~1h20). In order to make this work, 30 minutes after the slow train has left Cambridge North, the fast train…sits there for another 10 minutes, and then departs. By stopping at fewer stations on route, by the time it reaches London it has caught back up again to the 30-minute periodicity. On the return leg back north, the fast train arrives back 10 minutes earlier than its slow train equivalent would have done, and so it ends up sitting at Cambridge North…for yet another 10 minutes.

The other trains all tend to run at roughly regular half-hourly slots, into which it would normally be quite easy to add an EWR train every 15 minutes, because their symmetries are in sync. The Liverpool Street train’s timings, however, makes this simple task so much more difficult. As we will see, there are a few other minor quirks in the existing timetables too, all of which can be dealt with quite easily.

5.      Building more efficient timetables

Let us consider how this would work in terms of timetables. We’ve removed all at-grade crossings from the section, and Newmarket-bound trains have been moved onto a separate track. This should make our task relatively straightforward, as we have a single stretch of dual track between Cambridge Central and Cambridge North, with one train simply following the one in front. This is the most constrained section, so if it can be shown to work then everything else should fall into place.

We can start with the current timetable, and then see how the new services might look added in the ‘white space’. In this example, a normal weekday between 1pm and 2pm has been used – the differences from hour to hour are quite minor, so a proof-of-principle assessment needn’t replicate the analysis for every hour of every day. Instead of showing you lots of graphs and lists of timetables, I’ll stick with the ‘clock-face’ diagrams from the Appletown example above.

In the clock-face diagrams below, each service is represented by a 3-minute block, during which no other trains may depart. In the current timetable, we can see the two closest trains’ blocks adjacent to one another at xx:00 and xx03 on the northbound chart, and vast swathes of free ‘white space’ in both directions. These really aren’t very busy lines at the moment, with a train on each track between the stations just 30% of the time! But as you can see, the pairs of trains to Liverpool Street and (in the southbound direction) King’s Cross are not nicely symmetrical, with the Norwich train sitting where one might expect the second Liverpool Street train to be.

Now, if we attempt to add four hourly EWR trains to the existing timetable, it might look something like the figure below. At first glance all seems fine, but on closer inspection one of the four EWR trains has been slipped by a single minute from its ideal 15-minute-interval service, in order to avoid overlapping with an existing train’s 3-minute window. Although you might think this a very minor issue – arising from an existing timetable quirk which will long since have changed by the time EWR is built – as far as I can tell (going by Paul Sparrow’s above quote), it is precisely this which has led EWR Co to claim that the only solution is to build two more tracks through the north of Cambridge!

Nonetheless, sticking for now with the nearly symmetrical EWR trains added alongside the existing timetable, what other services might we need to make room for in the next few decades, to ensure our design is future-proof? A second Norwich train is an obvious one: Northbound it can fit exactly 30 minutes away from the existing train; southbound we have to offset it by a couple of minutes from its ideal position (one bit of broken symmetry forcing another one). After this, there is still space for another half-hourly service (e.g. trains to Wisbech) – and we’re still within the 75% resilience threshold stated in EWR’s requirements [Appendix B 5.7.2]. In both directions, there’s an oversized gap left in the hourly timetable for even a very slow freight train to get through, although this should be avoidable in most instances as freight could travel via Soham or Newmarket instead. We could even (not shown) double up the Birmingham service (which some hope could be extended to Leeds in future), by simply moving the northbound freight slot to ~xx.45.

The figure above illustrates how a northern approach – and a bunch of potential other future services – could work, without any changes to the existing timetable, without any signalling upgrades, and within the documented requirements of headways (3 minutes) and resilience (25% unused wiggle-room), without resorting to knocking down any buildings to build more tracks. However, what if we absolutely must have a symmetrically perfect-to-the-minute EWR service? Or, indeed, what if we would just like to tidy up the existing inefficient broken symmetries, and thus optimise the potential capacity through Cambridge’s only traffic-jam-avoiding cross-city travel corridor?

Northbound, the solution is incredibly simple. The fast train from Liverpool Street (which is about to sit at Cambridge North for more than half an hour) can just wait for a single minute at Cambridge station. That’s it.

Southbound, we have slightly more work to do. First, we align the time for which the King’s Cross trains dwell at Cambridge. I suspect the reason for this asymmetry in the current timetable comes from train-splitting, but with platforms at Ely and Waterbeach due to be extended, Network Rail themselves expect that procedure to be phased out in the coming years. Next, we have the fast Liverpool Street train (which leaves 10 minutes later than symmetry would like), and the existing Norwich train (which is then in the way). A neat, symmetry-restoring solution (as shown in the diagram below) is to push the Norwich train back by two minutes, and have the fast Liverpool Street train depart Cambridge North 10 minutes early, and dwell at Cambridge for 10 minutes (where it will be overtaken by the Norwich train, and then continue to Cambridge South roughly alongside the EWR service on the quad tracks). This would only be needed until two more (slow) trains between Cambridge Central and Liverpool Street are added, at which point both Cambridge North to Liverpool Street trains can become fast services.

The Norwich train, in ensuring regular hourly cycles, has a few minutes of wiggle-room to play with, but in a worst-case scenario it might miss its Stansted spot – in which case its lost slot could be taken up by an EWR train instead. A slightly less neat, but probably still workable and arguably less disruptive, solution (not shown) would have the existing Norwich train unaffected, the fast Liverpool Street train leaving just a minute or two earlier than it currently does, and the second Norwich train 32 minutes after (or 28 before) the existing one. These solutions involve very minor tweaks to the existing timetables, none of which should have any further knock-on effects on other services.

The above is, I would argue, a reasonable worst-case scenario for the 2040s, and is still acceptable within EWR Co’s given requirements. However, I suspect reality will be less constrained, as there are a variety of possible options which would each reduce the number of trains travelling through the busy corridor.

I would be slightly surprised if EWR needed four hourly trains, at least for the first decade or two. Even the busiest UK intercity services such as London to Bristol only have that many – as does Cambridge to King’s Cross (if we ignore the two slow trains which offer no utility for intercity passengers). EWR only serves relatively small towns and cities, and long-distance trips across the country which really aren’t that popular (and are mostly already quite doable via London, especially once Crossrail is operational). Half-hourly 8-carriage trains would have the same capacity as the proposed four hourly 4-carriage ones, and as EWR doesn’t really serve busy commuter routes east of Bletchley (where it’s expected that another 2/h will run from Oxford to the Milton Keynes  branch anyway), I suspect few would be very upset by a slightly less frequent service. Other than for a few select local routes (such as Cambridge and Bedford residents hopping on the train to Bletchley’s IKEA), EWR provides (at best) a modest improvement rather than a genuine transport revolution. Let’s just hope the trains have enough luggage space for flat-pack bookcases!

Next, it’s worth noting that Wisbech and King’s Lynn are a similar distance from Ely. The fast King’s Cross train, which currently splits at Cambridge (with only the front section continuing further north) could instead continue as one further north beyond the new Milton junction (where the track is less congested). It could then split in half at Ely (or the new Waterbeach station), with the front continuing to King’s Lynn, and the back departing for Wisbech three minutes later.

We could go further still. There are actually two routes from Norwich to Cambridge, of similar length. The route currently operating heads via Ely, but there’s another route via Diss and Bury St Edmunds. A direct train using the latter would either need to reverse in Stowmarket, or use a yet-to-be-built spur just north thereof. If a second Norwich train took this route, it would provide direct connections between Cambridge and Diss; Norwich and Bury St Edmunds. Moreover, while a large fraction of trips to Cambridge on the Norwich train involve people travelling the whole length of the line (with small numbers using intermediate stations such as Thetford), most Cambridge arrivals on the train from Ipswich come from Newmarket or Bury St Edmunds. Therefore, instead of having two Norwich trains via Thetford and two Ipswich ones via Bury, sending the second Norwich train via Bury may actually better serve passenger demand – as well as easing Ely’s congestion problems.

As a final potential tweak, it’s worth thinking about how to make best use of the station at Stansted Airport. Airports are the one location where direct trains are hugely advantageous, as swapping trains at intermediate stations is much more difficult if you have heavy luggage in tow (or are jetlagged!). However, the long single-track tunnel under the runway severely restricts how many trains can reach the airport. From Cambridge, two hourly trains make the journey – one from Norwich (Greater Anglia, class 755 diesel/electric hybrid trains) and one from Birmingham (Cross-Country, class 170 diesels). Could we use the advent of “Great British Railways” to shuffle rolling stock around, and thus find a way for these two trains to join together (at Ely or Waterbeach) and travel as one to Stansted? GA has a whole bunch of class 755s on local routes, a few of which could perhaps be swapped for CC’s class 170s – and I see no reason why the resulting 755 pair couldn’t couple together. This would then free up a Stansted slot, which (in a northern approach scenario) could enable an hourly direct EWR train from Oxford to Stansted!

If we put all of the above together in an optimistic scenario for the 2040s, we actually find ourselves with just a single extra train each hour compared with the current timetable (and even that is offset by an expected reduction in freight traffic, given a new spur south of Ely). I appreciate that at least some of the suggested optimisations might prove unworkable, but nonetheless it reinforces how many different options there are for ensuring there is absolutely no need to four-track through the north of Cambridge in a northern approach!

6.      Has this solved the problem?

The above examples show that there doesn’t appear to be any fundamental issue with scheduling all the trains we might possibly need in a way which avoids the need for 4-tracking. None of the tweaks suggested should cause any knock-on effects, so could be achieved while keeping all other timetables intact. Passengers boarding at Cambridge North towards Liverpool Street may see their southbound journey time increased by a few minutes – but I suspect that is an acceptable price to pay, given that a northern approach (unlike a southern one) would bring a regular EWR service to Cambridge North.

However, it is worth stressing that by 2030 (when EWR is expected to begin operation), the current franchises will be long gone: The current Greater Anglia franchise ends in 2025, and Great British Railways is about to completely shake up the way franchises work in the UK anyway. As such, so long as the *principle* of operation is sound, it’s not really worth worrying in fine detail what the precise timetables will look like at this stage – because they’ll all have changed by the time the line is built in any case.

It is, of course, worth noting that even in a southern approach scenario, the frequency of trains will increase – and symmetry-breaking carbuncles, which massively reduce line capacity, will need to be phased out regardless. The Liverpool Street asymmetry will probably be eliminated entirely within a few years anyway, as we expect to require a new pair of WAML trains which terminate at Cambridge Central. With four hourly trains in total, the Cambridge North pair will almost certainly both be designated as similarly ‘fast’, At which point the whole asymmetry issue disappears!

When designing new, it is always better to start as cleanly as possible, rather than adding yet more mess on top of existing inefficiencies. As the existing quirks can be easily tidied up – and will probably need to be resolved soon anyway – it would be wholly irrational to base major infrastructure design decisions around them. Nonetheless, given EWR Co’s claimed arguments, I thought it useful to illustrate how I think 4-tracking can be avoided both in principle, and in practice around the current timetable.

I haven’t considered in detail the extra services to London which Network Rail’s forecasts indicate will be needed by 2040. Most of these won’t pass north of Cambridge Central, so will not affect a northern approach. Looking at NR’s worst-case 2043 scenario, the only extra services needed north of Cambridge are one or two more Ely-KX trains (which could be the Wisbech trains, already factored into my model) and extending the two existing Liverpool Street trains beyond Cambridge North to terminate at Ely (which would remove the need for a 5^th platform at Cambridge North, but add no further constraint issues so long as the junction at Milton is grade-separated). On the other hand, they will add further pressure to a southern approach, which even with 4-tracking may find the south Cambridge corridor eventually becomes problematically congested – risking platform capacity issues at Cambridge Central from so many terminating trains, and perhaps a need to rebuild Shelford junction as grade-separated.

7.      Concluding remarks

EWR’s Paul Sparrow has stated that the existing timetable doesn’t “allow us to get 4 [EWR] trains an hour every 15 mins” for a northern approach. The analysis presented here appears to confirm that this is technically a correct statement, but – unless I am very much mistaken – only in the same kind of way that I can say I was Time Magazine’s Person of the Year in 2006: It’s only *just* correct, and likely very misleading for them to have used it to draw the conclusions they have.

EWR Co claim that the only solution to the above, in a ‘northern approach’ scenario, is to 4-track north of Cambridge station. I disagree. The immediate EWR issue can be solved by slipping a single hourly train by a single minute, and even if we are to future-proof to include the reasonable-worst-case scenario (including a variety of other future trains beyond EWR), only very minor tweaks to the existing timetable – which will almost certainly disappear soon anyway – are needed to make it work.

If my analysis turns out to be sound, then I do not understand why EWR Co decided that 4-tracking (with dozens of demolitions, new bridges etc) is considered a lesser evil. I accept that there may be additional complications of which we are not aware, and which render the proposals in this document unworkable – but if that is the case, then they absolutely must be able to explain to the public why this is the case, and cannot continue to simply state “our modelling says it is not possible” without offering a technical justification.

If the timetabling issue can be resolved and 4-tracking rendered unnecessary, the question of a northern vs southern approach becomes much more balanced. I will leave it to others to expound the pros and cons of each option, but advantages such as keeping more freight out of Cambridge, EWR regularly serving Cambridge North (and potentially also London and/or Stansted), the option of a station at Northstowe (offering direct links to Cambourne, Oxford and Addenbrookes), an easier way to get the Haverhill line back up and running (without it adding even more N-v-S imbalance to the numbers of trains running through the Cambridge corridor), and simplifying the process of electrification between Oxford and Cambridge by decoupling from the Eastern section (not due for upgrade until 2040-50)…are all worth thinking about very seriously indeed.

Why do we need to know about 4-tracking?

EWR Co’s decision to prioritise a southern approach into Cambridge primarily hinges on their assertion that a northern approach will require to be 4-tracked from the Milton junction with the existing West Anglia Main Line (WAML) into Cambridge. They also say that the existing Shepreth Branch Line between Hauxton Junction and Shepreth Junction (SBR) does not need to be 4-tracked but accept that this assessment needs further testing in later design stages.

It is vital that these matters are properly assessed now. To implement 4-tracking on either of these sections of existing lines would be extremely expensive. Major existing bridges would need to be modified or rebuilt and other infrastructure modified. If it is not resolved now, detailed design work would proceed based on a potentially faulty premise with the risk of locking the project into a solution that will lose valuable benefits of the alternative route.

To test EWR Co’s assessment, we have carried out our own evaluation of both their claims. We start by looking at the likely number of trains on the sections of the lines in question. We estimate the total number of trains likely to use the line including current traffic and probable growth for both passengers and freight. We discuss the likelihood of each of the scenarios occurring in practice and compare these rail traffic estimates with the capacity of the lines.

Summary

We found that EWR Co’s conclusion about the 4-tracking of the northern approach section NA2 is incorrect and that, for our estimate of the reasonable worst-case scenario, it does not need additional tracks. In fact, even with just traditional signalling there is likely to be sufficient capacity. With digital signalling, which is already specified by EWR Co, we expect there to be over-capacity.

Likewise, the southern approach section of the SBR line can probably take the reasonable worst-case passenger rail traffic estimate (which are based on Network Rail’s figures up to 2044). However, interfacing with the Thameslink services will impose severe restrictions on possible growth in traffic in excess of the figures shown due to anticipated expansion after 2044 as a result of the Ox-Cam Arc plans. The SBR line would not be able to accommodate EWR’s 6 trains per hour that are specified in the Project Wide Output Specification. The freight services estimated by Network Rail may not be able to be fully catered for because of limitations on the Newmarket line.

If EWR Co are wrong in their assertion that the SBR line does not need to be 4-tracked, the construction work that would be required exceeds the theoretical (but unrealistic) case for 4-tracking part of the northern approach.

Rail Traffic

The reasonable worst-case rail traffic scenarios for the northern and southern approaches to Cambridge are shown in Table 1. For the northern approach, we have considered the likely rail traffic on the critical section between Milton junction and Coldham’s Lane. All Newmarket trains would use platform 8 at Cambridge station enabling those services to avoid crossing movements between Cambridge Station and Coldham’s Lane. There is sufficient space in that length to increase the number of tracks should this be required.  

The assumptions and data sources used in compiling Table 1 are discussed below.  

• We have not assumed any long-term change in the pre-pandemic forecasts for growth and and ‘existing’ services due to impact of COVID, even though this is likely to reduce the amount of rail traffic in the short and perhaps the medium term.
• For consistency and independence, we have used Network Rail’s 2019 Cambridgeshire Corridor Study [Ref. 1] for assessing passenger and freight growth forecasts on the existing network for the two approaches. This will allow a fair operational comparison to be made between the approaches. We have resisted the temptation to consider in this assessment the beneficial effects (for a northern route) of an alternative future route via Newmarket for Norwich trains which has been suggested for many years, or the possibility of March trains splitting/joining other services at Ely. Combined, these could potentially reduce demand on the northern approach by 3tph. Nor have we considered the impact of a potential new Thameslink service between Cambridge and Maidstone East which may increase demand on the southern approach. Lastly, we consider that, if the Ox-Cam Arc proceeds as planned, there would be demand for even more Kings Cross services than mentioned in the Cambridgeshire Corridor Study with a timescale to 2044 only.
• Future demand for freight on EWR is taken from a scenario suggested by Network Rail in the EWR Co’s Technical Report 3.10.7.
• For a northern approach, EWR freight would be diverted northwards onto the WAML from Milton via a new grade-separated junction and then onto the main Felixstowe to Nuneaton freight line via a new chord just south of Ely. Any residual freight needing to pass through Cambridge would be planned out of peak hours.

Capacity

Basic ‘open-track’ Assessment

In addition to the basic number of trains per hour that could use an open railway line, the capacity is affected by a number of other factors, including timetabling, any crossing movements between tracks, whether the section of track is between junctions or between stations and the mixture of freight and passenger services. We start by looking at the open-line capacity and then go on to consider these other issues.

(i) Traditional signalling

Traditional signalling systems would allow for about 15 trains per hour with no other compounding factors. This is based on international standards (International Union of Railways or UIC which is the international rail transport industry organisation); a headway (the time between trains) of 3 minutes, as convention and a minimum required by EWR Co (Technical Report Appendix B 5.7.2); and a resilience ‘safety factor’ of 0.75 (UIC Code 406 for mixed traffic lines). The reasonable worst-case scenarios for both the northern and southern approaches as shown in Table 1 appears to be within the capacity of the existing twin track based on open track conditions and traditional signalling.

(ii) Digital Signalling

However, digital signalling will be used by EWR [Ref. 2] One of the benefits of digital signalling is the additional capacity that it provides. UIC have shown that on main railway lines, the capacity increase of ETCS level 2 (with ‘block’ lengths of 400m) over ETCS level 1 could be about 37% [Ref. 3]. This is supported by a statement by Network Rail quoting capacity increases ‘of up to 40%’ [Ref. 4]. These figures should be used with care, but they demonstrate that there would be a real and significant increase in capacity. Even using a capacity increase of half of these amounts, the number of trains per hour able to be accommodated could be increased to 18 by digital signalling without the need for 4-tracking,again without any complicating location-specific issues.

Importantly, digital signalling will also allow bi-directional running on tracks. This can provide significant flexibility, especially at stations, for reversing trains.

Other Factors in Capacity Assessment

There are several factors that could reduce actual capacity on a line, including whether clock-face timetabling is used (which is it on EWR) and the number of crossing of main lines radiating from London. Below we only consider those issues that are different between approaches.

(a) The Thameslink Effect and Slow Trains

Capacity reduction can arise when other train services use the section of track and the times of those services are not possible to change. This occurs for the southern approach where Thameslink services use the track. This leads to tight and sensitive interfaces with those services. The result is that the existing 6 passenger services are almost immutable because they form part of the complex Thameslink network and the East Coast Main Line traffic. The complexity of Thameslink can clearly be seen in the map[5] below, especially all the strands of services coming together in central London where it is planned to have 24tph crossing the Thames (which, incidentally, would not be possible without the capacity-enhancing benefits of digital signalling). EWR Co are very likely to be faced with predetermined ‘paths’ (planned slots for trains) at the Cambridge end of the route that do not mesh with their required paths elsewhere. This poses a serious risk to the effective capacity of this line, the outcome of which can only be resolved when timetabling of services is attempted. This is not the case for the northern approach.

If that were not sufficiently restrictive to the timetable, this line also has a slow service calling at all stations without a ‘loop’ or overtaking section for faster trains before they reach the East Coast Main Line near Hitchin. This means that timetabling for fast trains needs to allow for these slow trains, which will lead to greater gaps between trains. There are minimal such restrictions for a northern approach.

---

(b) Crossing Movements and Junctions

Crossing movements required across other tracks and junctions can also reduce the capacity of a line. In the case where junctions are flat (‘at-grade’), one train joining another line could have to wait if another train were using the section of track the first one wanted to use. ‘Grade-separated junctions’ involving ramps and bridges over lines significantly reduce this delay. The southern approach into Cambridge has two junctions (Hauxton and Shepreth Branch junctions) compared to just one for the northern approach.

The northern approach allows trains that are temporarily blocked by other trains to wait for a short time at Cambridge North station. This would allow following trains to also wait at Cambridge North (obviously on another platform) and so minimise any concertina-type delay. On the southern approach, conversely, delayed trains would have to wait in line between junctions, magnifying the impact of such a delay.

(c) Freight

Operationally, slower freight and more speedy passenger services do not mix well. Freight on a northern approach (via a new northbound chord at Milton onto the WAML and then onto the Felixstowe to Nuneaton line via another new chord south of Ely), means that freight would have no impact on EWR operations in the busy section between Milton junction and Coldham’s Lane junction. Conversely, freight and passenger traffic on a southern approach intermingle, potentially resulting in less overall capacity. If the line were eventually used for even slower moving, heavy-haul freight, such as building materials, this would only exacerbate the situation.

For a southern approach, the Cambridgeshire Corridor Study estimates that the Newmarket line, even with the anticipated dualling between Coldhams Lane junction and Teversham, could accommodate 1tph for freight during off-peak hours. Network Rail estimate the freight demand could be 1.3tph during an 18-hour window (i.e. including peak and peak hours), assuming the scenario suggested by Network Rail in the Technical Report 3.10.7. This implies that there could be insufficient capacity on a southern approach for the anticipated freight.

Another operational disadvantage of a southern approach is that because freight needs to pass through Cambridge and Cambridge South stations, the problem of platform provision in both locations would be exacerbated.

What Do EWR Co. Say?

Northern Approach

EWR Co have stated (Appendix F, 2.2.4) that the results of their analysis of a northern approach showed that there would be several conflicting movements between EWR and other services including:

• Trains towards Ely and eastbound EWR services conflicting where EWR services join the WAML; and
• Conflicts on various platforms at Cambridge station.

We find the first point hard to understand as there could be a grade-separated junction at Milton specifically to overcome this issue. We find their conclusion that the only remaining option is ‘to add two extra tracks to the WAML, making it a four-track railway between the new Milton junction and Cambridge station’ (Appendix F 2.2.10) curious when they could much less onerously provide a grade-separated junction at Milton than the 4-tracking option. Confusingly, they then state in Appendix F 2.2.12 that ‘a grade-separated junction is required where the EWR route joins the WAML…’.

The second point would be overcome by the construction of two more platforms at Cambridge station which EWR Co already confirm would have to be done (App F section 2.2.10). Interestingly, they fail to mention that the southern approach also needs two new platforms at Cambridge station (Technical Report 11.1.5 ). Indeed, as we will see later, there is more pressure on Cambridge station platforms from a southern approach because a freight service will need to pass through. CamBedRailRoad documents[1] provide a solution without the need for additional platforms at Cambridge.

Southern Approach

EWR Co state (Technical Report 11.4.1): ‘…it is most likely  that the SBR can remain as a twin track railway as there is sufficient existing capacity to be able to add the EWR services required to achieve the Project Objectives and leave spare capacity for an increase in services in the future.’ (Our underlining). So they have not confirmed this vital point before choosing a southerly route.

They go on to state: ‘The working assumption for the operational timetable will be assessed further in the next design phase to confirm that it is correct. The focus will be on timetable and performance modelling of the SBR to ensure that both the EWR and GTR services can run as required with suitable resilience to allow for delay, disruption, and updates to service patterns.’

They appear to acknowledge from the last statement that there could be a problem of interfacing with existing services but without being explicit about it. The Sponsor’s Requirements (App A, 5.3 and 5.4) oblige them, as far as practical, ‘to be resilient to any periods of poor performance on the wider network’ and to ‘isolate the wider network from any periods of poor performance on the Railway [EWR]’.

We consider that such a fundamental point as this should be resolved at this stage or, if it really cannot be resolved now, to base decisions on a risk-based approach (i.e. they should analyse the chances of success and the costs of each scenario) in making the choice between a northern or southern approach.

EWR Co mention that two more platforms are required at Cambridge station for a southern approach. This it is not necessarily just a case of removing some sidings to add an extra pair of platforms at the east of Cambridge station: the Royal Mail building really isn’t that far away, and may need to be removed or reduced to allow space for the extra switching tracks, particularly if another pair of side-by-side platforms are needed. Additionally, there is a very short distance between the platforms at Cambridge station and Hill’s Road bridge, in which space all the different platform tracks need to condense down to just four tracks. Indeed, that itself could become quite a bottleneck for a southern approach, as each of the various trains from London and Oxford terminating at Cambridge need to reverse & switch on their way back out.

Other Issues With Approaches

Although the above sections attempt to show whether the two approaches need to be 4-tracked, there are other arguments that need to be considered in the choice of route from a railway operations and cost viewpoint.

1. Symmetry

EWR Co state (Consultation Document page 52) that trains using a northern approach would need to terminate at Cambridge South station. Similarly, a southern approach would need to terminate at Cambridge North station since this area is also an important employment hub.

If EWR Co persist in their conviction that a northern approach would need to be 4-tracked south of Milton, a southern approach would also need to be 4-tracked to Cambridge North as it would carry at least the same amount of rail traffic. This means that, according to EWR Co’s logic, 4-tracking on this section would be needed whichever approach were adopted.

We fully understand some of the severe practical difficulties in providing this. But this underscores our assertion that the northern approach, by providing relatively easy access to all three stations in Cambridge, serves Cambridge’s employment needs better than does a southern approach.

2. Risk and Impact on Infrastructure

We have already referred to the risk of EWR Co being wrong in their future assessment of whether the Shepreth Branch line would need 4-tracking between Hauxton and Shepreth Branch junctions. If they were wrong, the structures that would require to be constructed or modified specifically for 4-tracking are:

• M11 crossing
• River Cam crossing
• Widened cutting E between Cam and A1301

The first two items are major construction works in their own right.

This list is in addition to those that are required even if 4-tracking is not required, including:

• Hauxton grade-separated junction,
• Harston level crossing changes,
• A1368 bridge rebuild,
• Hauxton Road level crossing changes,
• A1301 (rebuilt bridge likely if grade-separated junction at Shepreth Branch junction)
• Shepreth Branch grade-separated junction.

Clearly none of these 9 structures would require to be constructed if a northern approach were adopted.

Compare this to the two structures (the level crossing near Milton and the A14 crossing) that EWR Co state would be required if 4-tracking were required for a northern approach (which we, as detailed above, strongly refute). The other structures requiring modification (Fen Road, river Cam crossing, Newmarket Road and Mill Road) would need to be carried out anyway as a southern approach needs to terminate in Cambridge North station (see point 1 above). But all these works are less than those required for a southern approach.

This discrepancy in construction works between the approaches demonstrates EWR Co’s apparently sanguine approach to risk-based decision making.

3. Reversing Move at Cambridge

EWR Co have made much of the fact that a reversing move (i.e. the driver changing ends of the train) would be needed in Cambridge for onward eastbound traffic.

Two points are worth highlighting here. Firstly, it is not much of a penalty at all. A train approaching Cambridge from a southern approach would need to stop in the station for approximately 3 minutes before proceeding. For a train from a northern approach, the maximum time for the driver to change ends is about 5 minutes – a difference of some 2 minutes!

The second point is that onward eastbound passenger services are excluded from EWR Co’s remit. Although they need take such provision into account (the Sponsor’s Requirements 1.6 state that ‘Consideration should be given to the provision of or integration with services beyond the Oxford Cambridge sections…’), using this weak argument against a northern approach highlights their lack of strong arguments. It is especially rich when, in response to our queries about the problems that a southern approach would cause with freight east of Cambridge, EWR Co claim that it is not within their scope.

Conclusions

1. The northern approach would not need to be 4-tracked between Milton junction and Cambridge station for the predicted maximum amount of rail traffic including future expansion of services to Wisbech and Norwich. As a result of digital signalling, it could also cater for EWR Co’s aspiration of 6tph. EWR Co have appeared to ignore a grade-separated junction at Milton in their analysis of 4-tracking of this section.
2. It is not certain whether the southern approach between Hauxton and Shepreth Branch junctions (SBR line) can take any growth in passenger rail traffic beyond EWR’s 4tph until further timetabling work has been carried out by EWR Co. This is because of the severe constraints imposed by Thameslink and other services using the line. It would certainly not be able to provide capacity for the 6tph EWR services stated in the PWOS. This lack of flexibility to provide for growth is contrary to the Sponsor’s Requirements.
3. If EWR Co’s future assessment of the SBR line shows that they do require to 4-track it, the construction works required is likely to exceed the theoretical additional works required for 4-tracking the northern approach (as stated in conclusion 1, they are not actually required).
4. EWR Co have not been transparent about the adverse impact of freight using a southern approach on railway operations.

References

[1] Network Rail Cambridgeshire Corridor Study 2019.
[2] Programme-Wide Output Specification (Technical Report Appendix B 5.20.2) is ‘at least’ European Train Control System (ETCS) Level 2. The current signalling system in the general Cambridge area on existing track (i.e. in addition to the EWR track) is being upgraded to provide a platform ready for digital technologies and is planned to be complete by 2025, five years before the currently planned EWR completion date.
[3] ‘Influence of ETCS on line capacity – generic study’, UIC, Fig 24
[4] https://www.networkrail.co.uk/running-the-railway/railway-upgrade-plan/digital-railway/
[5] By Mvpo666 – Own work, CC BY-SA 4.0,
[6] CBRR Northern Approach Into Cambridge.