Should someone be taking a grip on new train procurement?

[35B]

For example the 80x has been selected as a relatively standard type for intercity trains. Newer units could be fitted with a battery pack instead of a GU and engines and interior fittings can evolve over time to reflect lessons and improvements. Nobody is arguing that they should all be exactly the same for decades, but an evolutionary approach is required, instead of a revolutionary one. (EMR should have just ordered 9 car 805s). The 805 appears to have been a pretty good example of a successful evolution of a trainset.

And the 810s are not. As importantly, we know there are material design defects in the 80x design, requiring significant rework, and which point towards the manufacturers trying to meet impossible procurement demands.

You'll never get down to one design for everything, but things like the tiny fleet of 12 397s ought to have been avoided, and interoperability between manufacturers should be prioritised.

Which is achieved by applying standards and insisting that they're consistently delivered to be authorised to run on the network.

== Doublepost prevention - post automatically merged: 26 Jun 2025 ==

Not necessarily vanity driven - just driven by the kind of short-term compartmentalised thinking that pervades our culture (and legislation) than come. Rather than come together and come up with a design that satisfies 80% of everyone’s requirements (and can be customised at surface level to satisfy a bit more), everyone insists on 99-100% at vastly higher expense because “I can only claim to have won if I get 100% of what I wanted and am visibly grinding everyone else’s faces in the dust”.

There's a separate question here, where procurement needs to balance precise specs (so you can actually run some decent acceptance testing) against flexibility to allow manufacturers to respond with the best value option.

 

[The exile]

The problem with a "standard" type is that by the time it exists everywhere the oldest ones will need replacing. Do you keep to the standard or do you upgrade to something incorporating the newest standards? The Southern Region was still building slam-door stock decades after automatic doors were a thing. Was that a good idea?

Not necessarily a bad one. Arguably what it actually did was continue to build medium and long distance slam door stock in line with the rest of the railway. Sliding doors for anything other than metro style units is a late 80s development. The first main line sliding doors stock on the SR was the PEP stock in the late 1970s - the first new metro stock they’d built for decades. That represented one of the breaks that are inevitable- what needs to be avoided are the micro fleets that won’t talk to each other. Common interoperability standards are the very minimum requirement.

 

[Snow1964]

There's a separate question here, where procurement needs to balance precise specs (so you can actually run some decent acceptance testing) against flexibility to allow manufacturers to respond with the best value option.

I would argue differently, precise specs are wrong way to go, much better to have max or mins (not both).

Seems much better to specify as (for example) fits C3 carriage gauge, can maintain 110mph with worn wheels, max power draw ?Mw etc rather than lots of specs that no one else has built a train to.

Let the manufacturers look at their box of parts, if they fit a switch or motor that is rated 20% higher because it is standard and cheaper (and is proven in service) let them, don't exclude it by having long lists of specs.

Remember the specs are only as good as the person who wrote them dreamt up, there is rarely any reason for some of them when they don't allow a better alternative that they didn't know was possible (but the manufacturer knows about, and can obtain quickly)


== Doublepost prevention - post automatically merged: 26 Jun 2025 ==

Presumably that is the tale of Aventra procurement, a not-quite-standard train, still bedevilled by the class 701 saga.

Maybe forgotten in all the standardisation talk is the rapid development of new technology, meaning the "standard" train you bought 5 years ago is now largely obsolete.
The demand is now for ETCS-ready bi/tri-mode trains with all the options available, not standardised DMUs/EMUs.
GBR integration also encourages GB silo-thinking, while no global manufacturer can afford to dedicate its output to the GB market.

There is too much change for vanity reasons, or introduced because it is new and latest thing (even if unproven), other railways in Europe are happy with 10 year framework agreements. Might end up with some components that are obsolete, but if there are 200 in a fleet, each using many of a part, then spares will be made, but might not be made for a tiny non standard fleet.

Much better to have standard parts on 10 years worth of trains, than lots of untried new era parts in small quantities. Helps maintenance too.

 

[Technologist]

However, the suppliers often obtain orders by threatening a politically damaging closure of a train manufacturing plant unless the government coughs up.

Orders are distributed based on threats of negative headlines, not based on merit.
We only need a few hundred vehicles per year on average, we only really have enough work for a single factory.
Maybe two at a push, I believe we now have four.

The UK market is not large enough for meaningful rolling stock competition, and I think it would be better if we admit it.

There is a solution, it's Soviet (and American) one.

In the cold war the Soviets had OKBs which were defence industrial engineering design outfits which normally bore the name of their leader (the most famous kept the name even after they retire or died) Mikoyan-Gurevich, Sukhoi, Tupolev, Antonov, Klimov, Yakovlev etc.

The design bureau designed the product, often with the help of TsAGI the state aircraft research organisation and then prototyped it at their own facilities. The production is then taken care of by state aircraft factory 1, 2, 3 etc.

The US has something similar where they have United States Air Force Plant 4, 6 and 42 where the production lines for their fighter aircraft are produced in a US government owned facility operated by a contractor.

Applied to railways we would have GBR central research, this would conduct research to help GBR be an intelligent customer writing specs around what is possible not just what existing manufacturers can make.

Then you would have the design bureaus who would design and prototype the trains using innovations that GBR research helped develop. GBR research would also reduce the cost of development by holding a large amount of IP centrally so different contractors don’t need to re-invent the wheel. It would also have the effect of extracting IP out of foreign manufacturers especially when they end up building their trains in the GBR train factories. The idea would that a relatively small start-up should be able to sell trains to GBR if they produce the best design.

The factory’s would then build the winning trains, whilst there isn’t direct competition here a factory is a lot easier for GBR to performance manage than a design house. GBR would also be able to buy foreign built trains if they wanted which would further incentivise the factory's to play ball.

 

[35B]

I would argue differently, precise specs are wrong way to go, much better to have max or mins (not both).

Seems much better to specify as (for example) fits C3 carriage gauge, can maintain 110mph with worn wheels, max power draw ?Mw etc rather than lots of specs that no one else has built a train to.

Let the manufacturers look at their box of parts, if they fit a switch or motor that is rated 20% higher because it is standard and cheaper (and is proven in service) let them, don't exclude it by having long lists of specs.

Remember the specs are only as good as the person who wrote them dreamt up, there is rarely any reason for some of them when they don't allow a better alternative that they didn't know was possible (but the manufacturer knows about, and can obtain quickly)

I largely agree - but that also means that acceptance is a lot harder to hold back, because you don't have something precise to measure against.

== Doublepost prevention - post automatically merged: 26 Jun 2025 ==

There is a solution, it's Soviet (and American) one.

In the cold war the Soviets had OKBs which were defence industrial engineering design outfits which normally bore the name of their leader (the most famous kept the name even after they retire or died) Mikoyan-Gurevich, Sukhoi, Tupolev, Antonov, Klimov, Yakovlev etc.

The design bureau designed the product, often with the help of TsAGI the state aircraft research organisation and then prototyped it at their own facilities. The production is then taken care of by state aircraft factory 1, 2, 3 etc.

The US has something similar where they have United States Air Force Plant 4, 6 and 42 where the production lines for their fighter aircraft are produced in a US government owned facility operated by a contractor.

Applied to railways we would have GBR central research, this would conduct research to help GBR be an intelligent customer writing specs around what is possible not just what existing manufacturers can make.

Then you would have the design bureaus who would design and prototype the trains using innovations that GBR research helped develop. GBR research would also reduce the cost of development by holding a large amount of IP centrally so different contractors don’t need to re-invent the wheel. It would also have the effect of extracting IP out of foreign manufacturers especially when they end up building their trains in the GBR train factories. The idea would that a relatively small start-up should be able to sell trains to GBR if they produce the best design.

The factory’s would then build the winning trains, whilst there isn’t direct competition here a factory is a lot easier for GBR to performance manage than a design house. GBR would also be able to buy foreign built trains if they wanted which would further incentivise the factory's to play ball.

The US and Soviet models are very different - and if you are going to advocate the Soviet model, you need to consider the success rate of those bureaux.

The same point goes for the idea of drawing in IP from foreign bidders - I can see little more calculated to deter bidding than such compulsory transfer when so many trains are designed on the basis of "platforms", benefiting buyers world wide by sharing development costs. That's before considering whether purchase only is the best way to buy trains, rather than build and run contracts.

 

[Technologist]

The US and Soviet models are very different - and if you are going to advocate the Soviet model, you need to consider the success rate of those bureaux.

The same point goes for the idea of drawing in IP from foreign bidders - I can see little more calculated to deter bidding than such compulsory transfer when so many trains are designed on the basis of "platforms", benefiting buyers world wide by sharing development costs. That's before considering whether purchase only is the best way to buy trains, rather than build and run contracts.

That's kind of the point, the design Bureaus could and did fail, not that many people lose work and those that do are likely soon to be working at another design bureau in the same region. The problem with the OKB structure in Soviet Russia was more related to how decisions were made in an unfree system. The advantage vs todays system is product worker jobs are not tied to product design so we don't end up procuring designs that aren't the best to avoid train factories closing.

Domesticating IP is precisely how virtually everyone who is now good at something went about it, I'm not suggesting that we necessarily demand design rights or patents but if you have people who have built Siemen's, Stadler and CAF trains in a career they are going to have a fair understanding of how their design solutions work.

The UK is the 9th largest passenger rail system in the work and probably has the potential to double or more in size, this should be enough to support a vibrant and innovative domestic rail industry.

 

[35B]

That's kind of the point, the design Bureaus could and did fail, not that many people lose work and those that do are likely soon to be working at another design bureau in the same region. The problem with the OKB structure in Soviet Russia was more related to how decisions were made in an unfree system. The advantage vs todays system is product worker jobs are not tied to product design so we don't end up procuring designs that aren't the best to avoid train factories closing.

Domesticating IP is precisely how virtually everyone who is now good at something went about it, I'm not suggesting that we necessarily demand design rights or patents but if you have people who have built Siemen's, Stadler and CAF trains in a career they are going to have a fair understanding of how their design solutions work.

The UK is the 9th largest passenger rail system in the work and probably has the potential to double or more in size, this should be enough to support a vibrant and innovative domestic rail industry.

Generally speaking, my view is that design and production are linked; separating the two breaks important feedback loops and denies the power of specialisation.

While I agree on the potential of U.K. rail, that doesn’t apply under the current procurement structure and changing production regime would do nothing for the sustainability of that.

 

[HSTEd]

I guess I am more extreme than most - I forsee a need for only two classes of train.

A pointy end fast train, probably battery assisted electric. Good to speeds of 320kph.
A gangwayed flat end train, battery assisted electric. Speed up to 170kph.

Both classes should be able to work in multiple (obviously the pointy end one can't use the gangway but you know).
Both classes should use the same charging adapter for static charging.
Both classes should be dual or tri voltage (25kV and 750V for sure, if 1500V is cheap, add it too)
Both classes should have all axles motored.
Both classes should have TGV M/Avelia Horizon style rapid refit interiors.

Both classes should offer at least partial level boarding from UK platforms. The slower class should proabbly be level boarding throughout.

If those desigsn can be developed then a combined order or 500 vehicles per year forever would set the railway up very well, even allowing for modest traffic growth.

 

[Brubulus]

And the 810s are not. As importantly, we know there are material design defects in the 80x design, requiring significant rework, and which point towards the manufacturers trying to meet impossible procurement demands.

Which is achieved by applying standards and insisting that they're consistently delivered to be authorised to run on the network.

While there was the cracking, that issue has now been sorted, unlike a certain manufacturer who has it as a perennial issue.

I guess I am more extreme than most - I forsee a need for only two classes of train.

A pointy end fast train, probably battery assisted electric. Good to speeds of 320kph.
A gangwayed flat end train, battery assisted electric. Speed up to 170kph.

Both classes should be able to work in multiple (obviously the pointy end one can't use the gangway but you know).
Both classes should use the same charging adapter for static charging.
Both classes should be dual or tri voltage (25kV and 750V for sure, if 1500V is cheap, add it too)
Both classes should have all axles motored.
Both classes should have TGV M/Avelia Horizon style rapid refit interiors.

Both classes should offer at least partial level boarding from UK platforms. The slower class should proabbly be level boarding throughout.

If those desigsn can be developed then a combined order or 500 vehicles per year forever would set the railway up very well, even allowing for modest traffic growth.

I agree with you here, except that the fast train should have 360kmh capability for HS2 and 180kmh is the maximum speed for gangwayed trains (any chance there could be a UK legislative change to make it 200) then you've got the perfect train for all but HS2.

Rapid refit interiors are a particularly good idea, since they enable a train to do regional, metro and long distance work over its life.

 

[35B]

A question. I’m in about 1985, and considering train designs for the forthcoming electrification to Weymouth. Do I:
1. Stick with more of the same (all now 10+ years old) Mk1 based units, or
2. Apply DC power packs to the 317s, or
3. Do something that reflects the premium nature of the route?

Another question. I’m in the mid 1990s, and considering new procurement to replace the “unsafe” Mk1 based units. Do I:
1. Buy more mk3 based units with EE507 motors; or
2. Invest in modern AC motored designs that will realise significant energy efficiency over the life of the train?

 

[HSTEd]

A question. I’m in about 1985, and considering train designs for the forthcoming electrification to Weymouth. Do I:
1. Stick with more of the same (all now 10+ years old) Mk1 based units, or
2. Apply DC power packs to the 317s, or
3. Do something that reflects the premium nature of the route?

If enough Mark 1 units are available to meet the requirement, you likely do that.
Otherwise you probably just build whatever EMU is in production.

As you know, we actually got something in between.

Another question. I’m in the mid 1990s, and considering new procurement to replace the “unsafe” Mk1 based units. Do I:
1. Buy more mk3 based units with EE507 motors; or
2. Invest in modern AC motored designs that will realise significant energy efficiency over the life of the train?

Energy efficiency savings aren't really that big of a deal in the 1990s.
Bulk electricity remained cheap and the insnae price escalation seen in the last decade could not reasonably have been predicted unless you were a prophet.

Whichever unit is cheaper to buy is likely to be cheaper overall.
There is a reason that even in the US DC motors clung on an awful long time - indeed I think some vendors still sell them as an option on newbuild locomotives today.

Today, the environment is very different. It is now possible to build a train that approaches the theoretical limits of train performance. Using electric car derived technology we can build a train which is traction limited on acceleration to very high speeds. Efficiency might creep up slightly, but there will be no more operational revolutions in train capability - we have more or less reached the fundamental limits of the technology.

 

[35B]

If enough Mark 1 units are available to meet the requirement, you likely do that.
Otherwise you probably just build whatever EMU is in production.

As you know, we actually got something in between.

Energy efficiency savings aren't really that big of a deal in the 1990s.
Bulk electricity remained cheap and the insnae price escalation seen in the last decade could not reasonably have been predicted unless you were a prophet.

Whichever unit is cheaper to buy is likely to be cheaper overall.
There is a reason that even in the US DC motors clung on an awful long time - indeed I think some vendors still sell them as an option on newbuild locomotives today.

Today, the environment is very different. It is now possible to build a train that approaches the theoretical limits of train performance. Using electric car derived technology we can build a train which is traction limited on acceleration to very high speeds. Efficiency might creep up slightly, but there will be no more operational revolutions in train capability - we have more or less reached the fundamental limits of the technology.

So when do you make the change? And what triggers that change? Nowhere else treats previous production as such a hard constraint on developing change.

 

[Brubulus]

So when do you make the change? And what triggers that change? Nowhere else treats previous production as such a hard constraint on developing change.

The 767 first flew in 1981. It is still in production today (for freight). I'm not saying the rail industry should follow aviation level timescales, but it's not like every industry redesigns it's products every few years.

 

[Zomboid]

There's no real need to be right up to the minute at all times - is there anything fundamentally inadequate today with products like 350/450s for regional EMU work?

A lot of it will be driven by the suppliers though, who are competing across a much wider market than just the UK, and they do work on those kinds of timescales. And the UK railway can only buy what's for sale.

The 767 is still in production because 1. Enough people want to buy them and 2. Boeing are willing to sell them.

We couldn't buy any new 450s now because Siemens won't sell them to us.

 

[The exile]

A question. I’m in about 1985, and considering train designs for the forthcoming electrification to Weymouth. Do I:
1. Stick with more of the same (all now 10+ years old) Mk1 based units, or
2. Apply DC power packs to the 317s, or
3. Do something that reflects the premium nature of the route

Given the fact that it is now approximately 30 years since true long-distance but non-IC stock was designed, you look at the requirements coming up (so that’s Weymouth, Kent Coast, arguably Sussex Coast, Kings Lynn, even possibly Norwich) and come up with something that will do for them all over an order period of the next (say) 20 years. OK -that’s an ac and a dc version, but there’s no inter working so dual voltage would be a waste (at 1980s costs)

 

[HSTEd]

So when do you make the change? And what triggers that change? Nowhere else treats previous production as such a hard constraint on developing change.

When a new design can deliver a substantial improvement in operational or economic performance or offer some major new capability.

In the mean time, standardisation of the fleet should be the driving ambition.

Two classes designed to minimise the training burden for staff should be the goal.

 

[35B]

There's no real need to be right up to the minute at all times - is there anything fundamentally inadequate today with products like 350/450s for regional EMU work?

A lot of it will be driven by the suppliers though, who are competing across a much wider market than just the UK, and they do work on those kinds of timescales. And the UK railway can only buy what's for sale.

The 767 is still in production because 1. Enough people want to buy them and 2. Boeing are willing to sell them.

We couldn't buy any new 450s now because Siemens won't sell them to us.

Well, the 350 now fails to meet crashworthiness standards for new trains, which may be a factor - independently of Siemens having moved on from the Desiro platform.

Meanwhile, Wiki tells me that the 767 has been made in a number of significantly different versions, with model upgrades taking place throughout the production lifespan of the aircraft. And the 737-Max tells us what happens when you try to keep something going for too long.

When a new design can deliver a substantial improvement in operational or economic performance or offer some major new capability.

In the mean time, standardisation of the fleet should be the driving ambition.

Two classes designed to minimise the training burden for staff should be the goal.

Those strike me as the worst possible reasons for selecting customer facing equipment. It takes us back to when my grandfather was told "we are not here for the benefit of passengers" when speaking to BR about improvements the City Corporation wanted to make.

 

[HSTEd]

Those strike me as the worst possible reasons for selecting customer facing equipment. It takes us back to when my grandfather was told "we are not here for the benefit of passengers" when speaking to BR about improvements the City Corporation wanted to make.

In an engineering sense, we will have reached the apex of what a steel wheel on steel rail train can do. After that, I don't see any other way for the engineering parts of a train to improve the passenger experience.

Fitting new interiors etc, sure. But the interior of the train is more or less inconsequential for the engineering of the train itself - that is why the TGV M expects to be able to replace the interior in 48 hours.
The train itself will become a box that fades into the background.

 

[dosxuk]

Fitting new interiors etc, sure. But the interior of the train is more or less inconsequential for the engineering of the train itself - that is why the TGV M expects to be able to replace the interior in 48 hours.
The train itself will become a box that fades into the background.

I'm sure that's what every rolling stock manufacturer has believed since mass production started. You only have to look at the work needed to fit retention toilets, or air conditioning, or to remove the shops from Voyagers, to realise that what seems obvious and complete at original design time is unlikely to remain true throughout the expected life span of modern stock.

 

[35B]

In an engineering sense, we will have reached the apex of what a steel wheel on steel rail train can do. After that, I don't see any other way for the engineering parts of a train to improve the passenger experience.

Fitting new interiors etc, sure. But the interior of the train is more or less inconsequential for the engineering of the train itself - that is why the TGV M expects to be able to replace the interior in 48 hours.
The train itself will become a box that fades into the background.

The idea that there can be no improvement on the best of today is for the birds. Building policy on that assumption is a guarantee of long run mediocrity or worse - and is part of what gave us such as the VEP.

Railways have always had multiple generations of traction on them, with a cascade process at work.

I was reminded recently, with a start, that the Desiros that I regard as "new" on SWR are all older than my son. He'll be 20 shortly.

 

[HSTEd]

The idea that there can be no improvement on the best of today is for the birds. Building policy on that assumption is a guarantee of long run mediocrity or worse - and is part of what gave us such as the VEP.

We now have the technology to build trains that approach the physical limits of what trains can do.

That was never possible before.
As an example, we will not be able to build trains that can accelerate faster 20 years from now than we can build today, unless we fundamentally abandon steel wheel on steel rail technology.

Railways have always had multiple generations of traction on them, with a cascade process at work.

And until privatisation this trend was slowly becoming less prominent, and indeed has continued to dissapear on other railways that were not so impacted.
For example, the subsurface lines of the London Underground now only have one train type.
30 years ago they had..... three (A, C and D stock I think?)?

The diversity of old fleets was caused by engineering limitations (fast units could not accelerate fast etc) or by rapid changes in technology.
Those limitations no longer appear to be binding in the same way they were before.

Indeed, a significant portion of the entire passenger rolling stock fleet in the UK is now composed of only a small number of classes. (especially 345, 700 etc)

I was reminded recently, with a start, that the Desiros that I regard as "new" on SWR are all older than my son. He'll be 20 shortly.

Well this is a good example.
20 years takes us from the start of Mark 1 construction to the PEP prototypes!

And yet modern trains have changed much less than that since the Desiros were built.

EDIT:
20 years from the PEP prototypes takes us to more or less the Class 323.
20 years from the Class 323 takes us to something like a Class 380

 

[35B]

We now have the technology to build trains that approach the physical limits of what trains can do.

That was never possible before.
As an example, we will not be able to build trains that can accelerate faster 20 years from now than we can build today, unless we fundamentally abandon steel wheel on steel rail technology.


And until privatisation this trend was slowly becoming less prominent, and indeed has continued to dissapear on other railways that were not so impacted.
For example, the subsurface lines of the London Underground now only have one train type.
30 years ago they had..... three (A, C and D stock I think?)?

The diversity of old fleets was caused by engineering limitations (fast units could not accelerate fast etc) or by rapid changes in technology.
Those limitations no longer appear to be binding in the same way they were before.

Indeed, a significant portion of the entire passenger rolling stock fleet in the UK is now composed of only a small number of classes. (especially 345, 700 etc)


Well this is a good example.
20 years takes us from the start of Mark 1 construction to the PEP prototypes!

And yet modern trains have changed much less than that since the Desiros were built.

EDIT:
20 years from the PEP prototypes takes us to more or less the Class 323.
20 years from the Class 323 takes us to something like a Class 380

And 30 years ago, the Networkers now heading for scrap were the latest best thing. There may not have been step changes, but there are still incremental advances - one of which is why the 380s are incompatible with the 350s.

 

[NCT]

Technology is advancing at a far faster rate than in the 80s. Modern trains are computers on wheels and software obsolescence is very quick. Wanting new rolling stock to be backward compatible is like insisting on buying a new computer on Windows XP.

Yes recent rolling stock introductions haven't been as smooth as they could be, but this is a global problem, and the problem can't be viewed through a British lens alone.

The reality is that rolling stock suppliers are now a relatively small number of global conglomerates. They are all trying to offer standardised but customisable platforms. It'll be the manufacturers driving the pace of evolutionary change, not some sort of New BR. British rolling stock will necessarily be global (European) platforms with British peculiarities:
- British loading gauge
- the 110mph speed category
- Gangway ends

That's not going to change unless we develop a British only, state directed, closed supply chain - that's for the birds even in Brexit Britain.

As far as suburban rolling stock is concerned we still have a long way to go before getting to the top of even just existing technology.
- Signalling and control - ETCS - there will always be teething problems as we roll out ETCS and have to have backward compatibility during the transitionary period
- Universal DOO/DCO - every time there will be a union battle, but the end goal has to be eliminating the 10-15 seconds before door open and after door close
- Automatic selective door opening as standard

How much of recent and current troubles are down to atypical levels of lack of standardisation anyway?
- Even in BR days we've always had 20m and 23m carriages. Having a 24m carriage version of the otherwise 26m IET shouldn't be rocket science.
- Having an EDMU with Class 222 performance on diesel is doing something that was never done before, but again that is really just an evolution on existing modularised components and shouldn't have been rocket science. That the requirement existed in the first place is more due to our chaotic electrification policy - not a rolling stock issue in itself.
- The 701 fiasco is a combination of build quality and multiple layers of luddites.

 

[Technologist]

The 767 first flew in 1981. It is still in production today (for freight). I'm not saying the rail industry should follow aviation level timescales, but it's not like every industry redesigns it's products every few years.

A few points on this:

The economics of aircraft basically boil down to fuel efficiency for most users who highly utilise their aircraft. To put this into context if I improve my engine by 3-5% fuel consumption over yours you literally couldn't give your engine away for free. Improve specific fuel consumption of the aircraft by 15-20% and you literally can't give the whole plane away. This means that for the next 5-10 years after entry into service aircraft engine manufacturers keep pretty large active development programs on the books improving the specific fuel consumption and durability. When engines get overhauled they will generally come back with the new standard.

Once past that point you will usually reach an architecture limit for the engine where improving thermodynamic performance won't actually help fuel efficiency much. After 5-10 years most aircraft then normally get stretched which normally coincides with the original engines reaching the plateau of performance. On most programmes at this stage after 10-15 years the development cost will be paid off and the airframer will start discounting the planes to allow them to be sold to freight users and charter airline who fly less hours and thus care more about purchase costs and maintenance costs than fuel efficiency.

That is pretty much it for the lifecycle of an airliner, after 20 years you are either going to be selling it as a freighter or if you are lucky like the 767 as a military aircraft, the military rarely fly their planes so don't care about fuel efficiency from an economic standpoint (they care about range). If you want to extend beyond a 20 year lifespan for the product you are look at re-engining the aircraft which typically means new wings, engines, control surfaces, undercarriage, cockpit and avionics essentially it's a new aircraft.

To justify a new plane you either need an new market segment or you need to be confident that you are going to get that 15-20% fuel efficiency bump that will justify the development risk/cost. However the key point is that the bits that do the work on aircraft are subject to continuous development development, the interiors tend to get refreshed pretty frequently as well, there is just relatively little scope for improving the fuselage of an airliner and that's the bit that tends to carry the name.

 

[D365]

I guess I am more extreme than most - I forsee a need for only two classes of train.

A pointy end fast train, probably battery assisted electric. Good to speeds of 320kph.
A gangwayed flat end train, battery assisted electric. Speed up to 170kph.

Both classes should be able to work in multiple (obviously the pointy end one can't use the gangway but you know).
Both classes should use the same charging adapter for static charging.
Both classes should be dual or tri voltage (25kV and 750V for sure, if 1500V is cheap, add it too)
Both classes should have all axles motored.
Both classes should have TGV M/Avelia Horizon style rapid refit interiors.

Both classes should offer at least partial level boarding from UK platforms. The slower class should proabbly be level boarding throughout.

No chance of achieving all of the above with full level boarding. Unless you envisage a Class 756 style solution - underfloor batteries will need to be accommodated using ”high floor” trailer vehicles.

 

[HSTEd]

No chance of achieving all of the above with full level boarding. Unless you envisage a Class 756 style solution - underfloor batteries will need to be accommodated using ”high floor” trailer vehicles.

Many continental units with low floors place equipment above the ceilings or in other nooks and crannies. It's worth noting that the BYD "Blade" Battery system uses prismatic cells that can be stacked into a pack only ~10cm in height.

Given that modern traction motor systems like those on the AGV can place the motors entirely inside the bogie frame, I think it is probably achievable.

 

[D365]

Many continental units with low floors place equipment above the ceilings or in other nooks and crannies. It's worth noting that the BYD "Blade" Battery system uses prismatic cells that can be stacked into a pack only ~10cm in height.

Which is achievable due to the greater loading gauge.

Roof-mounted equipment is being investigated here, too, but is very restricted in capability. Plus the figment of batteries at such a height requires additional consideration of the vehicle dynamics.

 

[Brubulus]

Which is achievable due to the greater loading gauge.

Roof-mounted equipment is being investigated here, too, but is very restricted in capability. Plus the figment of batteries at such a height requires additional consideration of the vehicle dynamics.

Remember UK platforms are generally about 20cm higher than European ones. This means there's effectively space for the equivalent of 2 BYD battery packs between the floor and the tracks, even on a low floor section. There isn't really a need for full level boarding, just for each unit to have 1 accessible coach, with lots of luggage, bike and wheelchair space.

 

[HSTEd]

Which is achievable due to the greater loading gauge.

Roof-mounted equipment is being investigated here, too, but is very restricted in capability. Plus the figment of batteries at such a height requires additional consideration of the vehicle dynamics.

As noted, level boarding in Europe requires an entrance at no higher than 760mm, in the UK it is 915mm.

So we still have room to stack batteries under the floor, even in a low floor section.

A 23m x 2.8m x 0.095m battery pack still has a volume of 6.1 cubic metres per vehicle.
It's probably more likely that Gallium Nitride power electronics would be fitted above the ceiling, rather than the batteries.

A battery assisted unit also doesn't need a large traction transformer, because it can use the batteries for acceleration and draw a lower but constant power from the transformer.
You might even attempt to fit one of those switching transformers that ABB was proposing.
EDIT:

I realise a battery under the floor probably only gets to be 15m long because it has to fit between the bogies. But still it comes out at around 4 cubic metres.

 

[D365]

Remember UK platforms are generally about 20cm higher than European ones. This means there's effectively space for the equivalent of 2 BYD battery packs between the floor and the tracks, even on a low floor section. There isn't really a need for full level boarding, just for each unit to have 1 accessible coach, with lots of luggage, bike and wheelchair space.

Exactly, and this is the approach that some train manufacturers are taking for future rolling stock designs.

I will reiterate that, unless a Class 756 style solution is utilised, the GB loading gauge cannot accommodate underfloor traction batteries on a fully motorised, low-floor vehicle.