Class 314’s to work the Gospel Oak - Barking route

[Hesperus]

I suppose technology has moved on, 20 years since the first ones entered service.

[jim.snowdon]

19 minutes ago, Hesperus said:

I suppose technology has moved on, 20 years since the first ones entered service.

Except that the problems are said to be with software, not hardware.

 

Jim 

[cheesysmith]

Part of the problem with these new trains is the fact they have moved on to multiplexing and the next generation of computer control. This saves mass and brings benefits in remote diagnostic and such, but it does require every bit of coding in the software to be safety verified, and changing one bit of code can cause problems with other bits, making the fault finding and testing a very only affairs.

[DY444]

11 hours ago, jim.snowdon said:

You would like to think that having got the Electrostar family firmly established, companies like Bombardier would capitalise on them rather than, as is seeming to be the case, starting a new family of trains from scratch. It's like change for its own sake, rather than necessity.

 

Jim

 

It was a necessity.  Iirc there are three generations of Electrostar platform (plus the 345 which is a bit of an Electrostar/Aventra hybrid) but you get to a point where the underlying architecture is such that you can't take it any further both because of its inherent limitations in operational terms and because of the increased risk of obsolescence during its life (bearing in mind that some equipment on the most recent Electrosars is the same as that on the original versions from nearly 20 years ago).  So Bombardier had to take the plunge with a new architecture at some point because the Electrostar is analogous to a crt in an HD flatscreen world.  It works but it's not what customers want.  Yes the new platform introduction has been painful so far but so was the introduction of the original Electrostars - they even had to provide extra free 357s because it was so bad.  By the time they got to the 387s they just worked and everybody forgets the shambolic start.

Edited April 10, 2019 by DY444

[Karhedron]

I agree, at some point you have to move product development forward rather than just incrementing existing designs. The tragic crashes of the new 737 MAX shows what can happen when you try to keep costs down by pushing an old design beyond its limits.

['CHARD]

12 hours ago, jim.snowdon said:

You would like to think that having got the Electrostar family firmly established, companies like Bombardier would capitalise on them rather than, as is seeming to be the case, starting a new family of trains from scratch. It's like change for its own sake, rather than necessity.

 

Jim

 

The problem is that the life operating cost of Electrostars is significantly greater than the Aventra family.  If Bombardier don't evolve their product platforms, they can't compete with Siemens, Hitachi etc., who are exploiting the new generation of technology.

[Joseph_Pestell]

1 hour ago, Karhedron said:

I agree, at some point you have to move product development forward rather than just incrementing existing designs. The tragic crashes of the new 737 MAX shows what can happen when you try to keep costs down by pushing an old design beyond its limits.

 

Those crashes are not about anything wrong with the basic 737 design. It's a simple wiring fault on the avionics.

[Joseph_Pestell]

On 07/04/2019 at 23:33, John M Upton said:

Stagecoach always believed in Guard operation so never specified it. 

 

Apart from having once operated Routemasters in Scotland, I don't think that there is any evidence that Stagecoach "believe" in operation with guards.

 

They would have gone DOO but Waterloo were very militant about it and Stagecoach gave in.

[Karhedron]

7 minutes ago, Joseph_Pestell said:

 

Those crashes are not about anything wrong with the basic 737 design. It's a simple wiring fault on the avionics.

 

No, it is more complex than that.  It stems back to the use of larger, more powerful engines being needed to loft the stretched frame. Because they did not want to redesign the whole airframe, they stuck the engines where they had always been on a 737. But this affected the balance of the aircraft and led to the Boeing introducing the flawed MCAS system. Far from being a simple wiring fault, it was a cascade of failures resulting from an attempt to upgrade rather than replace the basic 737 airframe. This was then compounded by making basic safety features that would have spotted the error an "optional upgrade" in an attempt to cut the price.

 

If you want a detailed breakdown of what went wrong, there is a good summary here. It shows that the whole problem is much more complex than just a simple wiring fault.

 

https://www.seattletimes.com/business/boeing-aerospace/failed-certification-faa-missed-safety-issues-in-the-737-max-system-implicated-in-the-lion-air-crash/?fbclid=IwAR3KyBV1u0-N5Ojr1IUzUvCKPaSRStYvCDyD-D2j1kQlWucyjWT60qfmtdM

['CHARD]

It's got Edsel or more accurately Pinto written all over it, this tragic mess, in PR terms.

[Edwin_m]

1 hour ago, 'CHARD said:

 

The problem is that the life operating cost of Electrostars is significantly greater than the Aventra family.  If Bombardier don't evolve their product platforms, they can't compete with Siemens, Hitachi etc., who are exploiting the new generation of technology.

Wasn't there also a new version of crashworthiness standards that would have required a major re-design of the cab if they had wanted to produce more Electrostars? 

['CHARD]

5 minutes ago, Edwin_m said:

Wasn't there also a new version of crashworthiness standards that would have required a major re-design of the cab if they had wanted to produce more Electrostars? 

 

Yes indeed.  The last build of 387s - I think, ordered speculatively by Porterbrook and now leased to GWR - were the last gangwayed units new on UK metals to the outgoing crashworthiness standard.  I suspect that the new ScotRail AT200s (class 385?) are the first constructed to the new regs, and that wasn't a startling success....

Edited April 10, 2019 by 'CHARD
brain fade

[St. Simon]

15 hours ago, jim.snowdon said:

Except that the problems are said to be with software, not hardware.

 

Jim 

 

The software on the 345s is the old Electrostar Software, not new software.

 

Simon

['CHARD]

41 minutes ago, St. Simon said:

 

The software on the 345s is the old Electrostar Software, not new software.

 

Simon

 

As I understand it, the Class 345 fleet as delivered (at least up to February), has seven upgrades/ iterations of software installed.

[jim.snowdon]

3 hours ago, 'CHARD said:

 

As I understand it, the Class 345 fleet as delivered (at least up to February), has seven upgrades/ iterations of software installed.

I rest my case. The Electrostar is a mature design by now, so why change its software? Or doesn't the old adage of "if it ain't broke don't fix it " apply to software?

 

Jim 

[mdvle]

1 hour ago, jim.snowdon said:

I rest my case. The Electrostar is a mature design by now, so why change its software? Or doesn't the old adage of "if it ain't broke don't fix it " apply to software?

 

Jim 

 

Part of the problem is that the hardware underneath is no longer available (or at least not for the 20+ years starting from bid time) that the software runs on.

 

Then add in that the owners/operators want more features from the hardware/software and offering a "mature" platform is neither cost effective nor liable to produce a winning bid.

[St. Simon]

2 hours ago, jim.snowdon said:

I rest my case. The Electrostar is a mature design by now, so why change its software? Or doesn't the old adage of "if it ain't broke don't fix it " apply to software?

 

Jim 

 

If we used the adage "if it ain't broke, don't fix it", we'd still be using the 'Rocket' as motive power.

 

Simon

[phil-b259]

On 10/04/2019 at 20:22, mdvle said:

 

Part of the problem is that the hardware underneath is no longer available (or at least not for the 20+ years starting from bid time) that the software runs on.

 

 

Exactly!

 

Just try getting hold of a Pentium 3 or Processor these days to run Windows 98 on (yes thats the software the Electrostars were fitted with initially although later builds moved on to Windows XP)

 

Signalling designers faced exactly the same problem - BR SSI may well be a nice 'mature' product, but (i) the software simply will not work on the much faster computer chips around these days but (ii) NOBODY makes computer chips to a 1987 specification only more!

 

Around a decade ago it therefore became necessary for the big signalling suppliers to produce whole new generations of interlockings whose only link with the BR SSI was that the protocols used for the data links were the same so the old BR trackside modules could be retained. Otherwise it was new computing hardware and new software to match. Even these may well end up obsolete in 20 years thanks to the rapid advances in the computer world.

 

 

Edited April 11, 2019 by phil-b259

[roythebus]

Computer software and hardware is obsolete before it's even reached the production line. that's the whole problem with it. At least relays and switches still do the same basic function they've always done.

[jim.snowdon]

Computer software exists at several levels. What actually runs on the processor is the code, and is specific to that family of processors. Above that is the program, as a set of logic instructions written in a programming language and essentially processor agnostic, and above that the program as a set of logical instructions. 

 

Jim 

[mdvle]

The problem is that things like trains are only sort of using a PC.

 

In addition to the problem of the software making assumptions that later generations of hardware make false, there is the problem of all the interface hardware that connects the "PC" to the rest of the train, and more than anything it is that hardware that has become obsolete (the hardware isn't make anymore, the drivers that allow the OS to speak to it haven't been updated to new OS versions and never will, etc.).

 

Then, on the software and hardware side, because of the nature of what these do - brakes, etc - means the specification is tightly locked down and tested.  What this means is that the "system" is only certified for use with specific hardware - down to a specific CPU, specific hardware, specific memory (not just size, but down to the manufacturer and performance specs), specific version of the OS, specific version of the compiler, etc.  If you are forced to change any of those items, then you need to restart the testing/certification process...

[RANGERS]

On 10/04/2019 at 21:12, St. Simon said:

 

If we used the adage "if it ain't broke, don't fix it", we'd still be using the 'Rocket' as motive power.

 

Simon

If we were, Gospel Oak to Barking might have some more trains...

[david.hill64]

12 hours ago, mdvle said:

Then, on the software and hardware side, because of the nature of what these do - brakes, etc - means the specification is tightly locked down and tested.  What this means is that the "system" is only certified for use with specific hardware - down to a specific CPU, specific hardware, specific memory (not just size, but down to the manufacturer and performance specs), specific version of the OS, specific version of the compiler, etc.  If you are forced to change any of those items, then you need to restart the testing/certification process...

This is precisely why manufacturers are extremely reluctant to change platforms unless they must.

Typically a microprocessor supplier will issue a notice of impending withdrawal of components and the manufacturer will take a view on likely orders prior to acceptance of the next generation equipment and buy in stocks of the soon to be obsolete equipment so that production can continue.

 

Whereas Microsoft/Apple/Google can get away with releasing operating systems with bugs, suppliers of safety critical systems cannot and the verification and validation activities required prior to release are onerous. Even after extensive lab testing it is not unusual to find that field testing throws up some googly. Hence different software versions, each of which is locked to a particular platform and each of which has to be approved.

 

The trend now is to move towards standard hardware platforms for multiple applications. So for example the hardware for the mainline ETCS signalling system may find an application in a metro signalling system.

 

Remember too that evaluation criteria for systems has changed. Despite protestations to the contrary, first cost was always a winner in previous times. Nowadays with availability targets more important and maintenance being the responsibility of the supplier, it makes more sense to add redundancy into equipment. I suspect that in Bombardier's case the Aventra family will have considerably more equipment than the Electrostars. Once the initial reliability growth period is over the availability will be better. Trouble is that with more equipment the reliability growth period will be longer.

[DY444]

On 10/04/2019 at 19:09, jim.snowdon said:

I rest my case. The Electrostar is a mature design by now, so why change its software? Or doesn't the old adage of "if it ain't broke don't fix it " apply to software?

 

Jim 

 

It does but that isn't the point.  The existing train software did not support what was required and therefore had to change.  The signalling system selected for the central section of Crossrail is not found anywhere else that Electrostars have operated.  Software to operate with that signalling system had to be developed and integrated with the existing train software.  Aside from the inevitable bugs in a complex software upgrade, the signalling system software itself is going through revisions to deal with issues that have arisen (principally with the interfaces to the conventional signalling) and is thus something of a moving target.  

Edited April 12, 2019 by DY444

[jim.snowdon]

I may be looking at this from a more traditional viewpoint, but signalling software - the modern equivalent of the auto-driver box - is a nominally stand alone system which drives the train by providing alternate inputs to those that would have come from the driver's controls. As such it isn't part of the train 's core software but a system feeding into it, through an interface.

 

Jim