Junction Relief

To amplify that point, there was some very significant work carried out to the Aldgate Triangle before the second world war, to the extent of providing a new tunnel for the District side of the triangle, and yet they did not take the opportunity to provide any grade separated junctions. I cannot imagine they never considered the idea and so it must have been rejected on the grounds of cost and or complexity.

The name of the game will change with resignalling (if/when it comes) taking a lot more of the decision making away from humans and allowing a faster, closer approach to a converging junction.

While I fully understand your point of view I have to disagree with your analysis.

When the Aldgate area resignalling was commissioned I was in Farrdingdon signal cabin commissioning the SPTs from Aldgate to King's Cross. It was a long night and even at Farringdon I could hear the Signal Engineer (Joe Kershaw) at Aldgate over the phone telling his supervisors what their futures might hold if they didn't get things sorted ASAP. Joe had a way with words! of course there wasn't a great deal that they could do because the problems were design and software issues. When the Aldagte area regulator at Farrindgon pushed a button to clear a route once trains were running in the morning he was querying why signals weren't clearing and then, when they had, why the trains were stationary. The reason was the way in which things were done, AIUI then the regulator pushed a button sending a command to Baker Street computer, which processed it and sent a command to Aldgate to clear the route. Basically it could take up to a full minute from route push to route clear.

Computer control and manual control are effectively the same thing, what is different is the signal operator! The experienced signal operator able to operate a site efficiently was unfortunately not the computer programmer, had the computer been able to replicate the actions of an efficient signal operator the computer would win every time simply because it has a better attention span and faster processing.

The Aldgate problem IMHO was two-fold, firstly the computer was programmed by a programmer interpreting what a signal operator would be expected to do and secondly a slow communications interface between sites. I believe that in any future resignalling neither of these problems will exist, it really all boils down to having the correct processing algorithms in place and at suitable locations.


It is true to say that a human brain is a computer and still far cleverer than the manufactured item but the latter is simply so much faster and doesn't have to stop for lunch etc but it is only as good as its programmer so if you want a computer to do a better job than a human make it possible for the particular human to be able to programme it, at present AFAIAA this is still at the monkey see monkey do stage but over time it will progress, computers have for some years been doing things that humans simply cannot do or might be able to do if they lived long enough!

Yep you just can't beat a decent set of EP points, even when the air pressure is quite low a good set will still throw fast. I've never had any dealings with electric points but my understanding is that they take time to 'wind' across. I suspect there may be more than one reason for a move towards them, not least of which will be the complaints from passengers and staff alike about the noise associated with air exhaust from the throw of EP points. I laughed my head off some years ago when we had to 'silence' a D valve at Acton Town because the exhaust from it frightened a member of staff walking the staff pathway from the station to Bollo House!

Wasn't there a distant plan by the Met before nationalisation to tunnel down the western bay at Baker St to grade separate the northbound line north? I recall this because something I read mentioned that the new planned lift shafts would impinge on what the plans would have required. ...but with that, you could have a non-conflicting centre terminating road, and knocking through the other bay (and in all likelihood, turning the current pair of through roads into bays) would have made Baker Street a whole lot more efficient.

The only plan I have read about was to eliminate the two-track bottleneck between Finchley Road and Baker Street by building a new line under ground from Finchley Road to west of Edgware Road. The Met got as far as rebuilding Edgware Road, but then was swallowed by LT which preferred to extend the Bakerloo.

This plan is why Edgware Road has the extra platforms that are now used to turn District and Circle trains.

I thought too it was to cater for the 'other' line up the Met. that appeared in the Railway Mag. article at the time of the resignalling.



Not really, AIUI. You've still got a motor and leadscrew effort - none of this big-piston-goes-whoosh-CLONK sort of speed.

Disagree again I'm afraid. First off, the communication to site is as fast as all the other sites, and as fast as you generally get anywhere when controlling remote equipment. There has never been a problem with any "lag."

Second, the current programming (in terms of what it does and what it decides) has a great deal of input from me. A few years ago I, along with a couple of the people who wrote the software for the site, ripped the original design and routing specs apart. The site now does things slightly better than it used to, and makes certain decisions faster. Some of the decisions it makes have been re-written, so instead of it doing "x because y is happening" it now tends to do "y now, because y can happen now and will help x" if you see what I mean.

But Aldgate has so many different scenarios, with so many different outcomes, you would need a pretty decent computer to be able to manage each and every decision as smoothly as a human. That kind of technology is not within the budget of our company, and as such will never been employed.

Which leads us back to the original point that computer control - as far as we'll ever have it - being worse than human control at some locations.

Really? OTTOMH I can only think of 6 or so that really matter - of course it may look different to an operator rather than a designer.

Mmm! Looking back over 28 years I would suggest that the performance of signal operators varied greatly. As in all things no two are exactly alike and like controllers some are definitely a cut above the rest when it matters.

I never had the opportunity to work on the signal computer section but I did maintain the computer controlled signalling at the east end of the Picc and the computer programme machines at Heathrow.
IMO perhaps the biggest problem in computer control of LT signalling is the method of operation. Even with a decent signal operator entering commands at the terminal the time taken to move a lever always seemed inordinately slow. I suspect that was due to the way in which everything was proved before clearing the route. Programme machines did not suffer the same problem so one might be forgiven for thinking that 1950s/60s discrete electronic technology was better than 1970s/1980s/1990s computer technology but this is like mixing apples and oranges.
The P/MC is just running a timetable with fixed programme and a very limited and finite amount of decision making, i.e. storage, whereas the computer is supposed to be doing much more. The P/Mc relies upon both safety and non-safety interlocking but AFAIAA the computer also relies upon software interlocking. This is a fundamental difference, the P/Mc actually being closer in nature to a signal operator than a signalling computer, the signal operator relying only upon the safety and non-safety interlocking to inhibit bad decision making.
If the signalling computer was programmed only to issue commands and to respond to events as a signal operator does and leave the interlocking to the safety and non-safety signalling equipment it would no doubt give a much better performance. My understanding is that to date LT/LU signalling computers have also become much of the non-safety circuitry and if that is so then perhaps what is required is a 'division of labour' using true parallel processing.

FWIW, the electric points at Alton on the Mid Hants Railway throw, lock and prove in under three seconds.