Post by CSLR on Feb 10, 2006 19:43:46 GMT
This post follows on from previous comments. While it can be read as a stand-alone piece, it might make more sense if you first look at the post Gate Stock Withdrawal - The Great Leap Backwards? at:-
districtdave.proboards39.com/index.cgi?board=History&action=display&thread=1139590164
When an automatic system is designed to replace a manual system, it can be in the form of replication, simplification or enhancement. In other words, the programmer can:-
1. Copy the manual system ‘as is’
2. Copy the manual system and add a few bells and whistles
3. Copy the basic idea, but leave out some of the frills
In complex systems, there is often a mixture of all three types in the final configuration and these will impact on each other. This may be fine in a controlled environment, but in some cases (including London’s Underground) there is an un-automated wild-card that will always be there. While you may fully automate the actual operation of the trains and stations, you cannot automate the travelling public. The very best you can hope for is some form of control. But control relies on compliance, and with the huge numbers of people that we are looking at there will always be those who will act unpredictably. Effectively, the more you automate, the bigger this type of problem becomes.
So, we design exceptionally efficient smart new trains that think for themselves and we design integrated signalling (visible or not) to maximise the operating potential. Then we allow the trains to stop at stations and we let loose hordes of unpredictable passengers that do not fall within the control of the automated systems.
Other problems arise when we allow people to see things that are automated. For a few this is too great a challenge and they are tempted to demonstrate their superiority by vaulting ticket barriers or obstructing train doors. Hey! Train doors. I have talked about that subject before. Let me now flash back to the gate-stock that I mentioned in the earlier post. Looked at in computer terms, what you then had was a human interface at each boarding and exit point that was able to interact with what was happening at that specific location at that precise moment (let us call that system ‘A’). When gatemen were replaced, the system was simplified to allow one person to control all of the doors (we will call that system ’B’). Today, we still effectively operate the simplified system B and all future automation that I have heard about suggests that we cannot easily revert to A. We have therefore accepted a compromise whereby we reject a better passenger interface and better passenger control for the sake of simplicity. Nevertheless, we cannot blame the computer programmers because of a decision that was made during an earlier attempt at automation; they are simply copying what they have inherited. Looking at it realistically, it would now be possible to design a computer system that would operate the doors more efficiently, but it would be prohibitively expensive and add another layer for potential failure.
At this point we should remember the butterfly effect. This is a theory that any action taken will eventually result in something happening elsewhere, often in an unpredictable way. While a computer model of a machine can suggest what it will do and how it will do it, it cannot predict exactly what Joe public will do to screw it up. We have become so reliant on computers that we often forget that they cannot think; they simply make calculations based on the facts that they are given. This was clearly demonstrated by computers that were unable to predict all of the problems that arose with the introduction of ATO on the Central Line. It happened because the computers did not have a large enough accumulation of knowledge of the type needed to react usefully. Put simply, it is the computer equivalent of a learning curve and it is why computers are sometimes better at solving old problems than new ones.
There is no doubt that LU has a wealth of experience of ATO in general, but putting that experience into operation across the entire system with new rolling stock, complex timetables, a mass of junctions, antiquated tunnels, etc will be no easy task.
I personally look forward to observing any attempt to introduce ATO on the existing system. An automated District and Metropolitan could offer interesting challenges, particularly if the service goes belly-up. If the staff cannot always restrain themselves, just imagine the day when we see green and maroon computers fighting it out in some form of high-tech inter-line rivalry?
With all these thoughts in mind, let us ask that string of questions again:-
1. Will system wide ATO be more reliable? There are three answers to this question.
(a) To physically move the train (by replacing a T/O with a machine), the answer is “Probably”, although it might help if the next generation of onboard control equipment learns a little more about leaves and rain.
(b) Reliability of equipment in an automated system is a separate issue. There is certainly more to go wrong on an ATO system. It is also a fact that the process of automation does not in itself improve the reliability of the electronic, electrical or mechanical equipment - these remain as issues that must be addressed separately
(c) The matter of ATO reliability when interacting with passengers on a high-density service certainly needs more thought. Computer controlled machines tend not to react well with humans (consider a room full of senior citizens and a video recorder).
2. Will it allow for faster journey times?
While automation could result in higher speeds, there is a finite limit to what can be achieved in existing tunnels. In theory you could bank the track round a curve, but you cannot do much with a dip. The fastest speed is not determined by the performance of the train, it is governed by what may be operated safely and comfortably – this also applies to braking. It is true that ATO trains have been allowed to drive at speeds that manual operators would never have been permitted to, but it ended in tears when the trains tried to show off. Now the computers have had the programming equivalent of a slap on the wrist and their speeds have been reduced. There may therefore be some extra speed, but it will probably not be as much as had been hoped for, at least not in the immediate future.
Also, as ATO goes system wide there is talk of ‘simplifying routes’. If implemented (and this seems to be a plausible option), it will result in more passenger changes and therefore a greater number of people in the system at any time. Route simplification would also introduce the secondary (though temporary) problem of disruption caused by the essential redevelopment of interchanges.
The bottom line is; the trains might travel faster, but the passengers might not.
3. Will we get higher line capacity?
Remember the maxim, ‘a motorway is the quickest route between two traffic jams’; you may gain a little extra time between stations, but I cannot see too many extra trains being packed in. We must still consider safety, so if you add more trains and reduce the distances between them you may have to reduce the line speed (see 2).
The argument that the new rolling stock offers greater capacity is not relevant as its design is not reliant on ATO.
Before we move on, let us put these last two points together. Simplified routes = more changes = longer journey times for passengers = more people in the system for longer = more congestion. Yup, we are going to need all the extra capacity we can get.
4. Will it be safer?
Well, you do eradicate the possibility of much human error, but you will not rule out the possibility of a programming error, of an equipment breakdown or of a type of leaf on the track that is not recognised by an onboard computer.
As newly constructed lines, DLR and the Victoria Line were able to run a real-time. experimentally service for an extended period prior to opening. The Central did not have that luxury and it showed.
To sum up: The changes that will occur as a result of the introduction of ATO may be noticeable in ways that were not really intended. While system wide ATO will be expected to enhance the service, there will be instances where it will not. That is a fact that the designers of the system will be well aware of; and it does not need a computer to work it out. As in any pioneering system, there will be several steps forward, but do not be surprised to see a few steps backwards as well.
districtdave.proboards39.com/index.cgi?board=History&action=display&thread=1139590164
When an automatic system is designed to replace a manual system, it can be in the form of replication, simplification or enhancement. In other words, the programmer can:-
1. Copy the manual system ‘as is’
2. Copy the manual system and add a few bells and whistles
3. Copy the basic idea, but leave out some of the frills
In complex systems, there is often a mixture of all three types in the final configuration and these will impact on each other. This may be fine in a controlled environment, but in some cases (including London’s Underground) there is an un-automated wild-card that will always be there. While you may fully automate the actual operation of the trains and stations, you cannot automate the travelling public. The very best you can hope for is some form of control. But control relies on compliance, and with the huge numbers of people that we are looking at there will always be those who will act unpredictably. Effectively, the more you automate, the bigger this type of problem becomes.
So, we design exceptionally efficient smart new trains that think for themselves and we design integrated signalling (visible or not) to maximise the operating potential. Then we allow the trains to stop at stations and we let loose hordes of unpredictable passengers that do not fall within the control of the automated systems.
Other problems arise when we allow people to see things that are automated. For a few this is too great a challenge and they are tempted to demonstrate their superiority by vaulting ticket barriers or obstructing train doors. Hey! Train doors. I have talked about that subject before. Let me now flash back to the gate-stock that I mentioned in the earlier post. Looked at in computer terms, what you then had was a human interface at each boarding and exit point that was able to interact with what was happening at that specific location at that precise moment (let us call that system ‘A’). When gatemen were replaced, the system was simplified to allow one person to control all of the doors (we will call that system ’B’). Today, we still effectively operate the simplified system B and all future automation that I have heard about suggests that we cannot easily revert to A. We have therefore accepted a compromise whereby we reject a better passenger interface and better passenger control for the sake of simplicity. Nevertheless, we cannot blame the computer programmers because of a decision that was made during an earlier attempt at automation; they are simply copying what they have inherited. Looking at it realistically, it would now be possible to design a computer system that would operate the doors more efficiently, but it would be prohibitively expensive and add another layer for potential failure.
At this point we should remember the butterfly effect. This is a theory that any action taken will eventually result in something happening elsewhere, often in an unpredictable way. While a computer model of a machine can suggest what it will do and how it will do it, it cannot predict exactly what Joe public will do to screw it up. We have become so reliant on computers that we often forget that they cannot think; they simply make calculations based on the facts that they are given. This was clearly demonstrated by computers that were unable to predict all of the problems that arose with the introduction of ATO on the Central Line. It happened because the computers did not have a large enough accumulation of knowledge of the type needed to react usefully. Put simply, it is the computer equivalent of a learning curve and it is why computers are sometimes better at solving old problems than new ones.
There is no doubt that LU has a wealth of experience of ATO in general, but putting that experience into operation across the entire system with new rolling stock, complex timetables, a mass of junctions, antiquated tunnels, etc will be no easy task.
I personally look forward to observing any attempt to introduce ATO on the existing system. An automated District and Metropolitan could offer interesting challenges, particularly if the service goes belly-up. If the staff cannot always restrain themselves, just imagine the day when we see green and maroon computers fighting it out in some form of high-tech inter-line rivalry?
With all these thoughts in mind, let us ask that string of questions again:-
1. Will system wide ATO be more reliable? There are three answers to this question.
(a) To physically move the train (by replacing a T/O with a machine), the answer is “Probably”, although it might help if the next generation of onboard control equipment learns a little more about leaves and rain.
(b) Reliability of equipment in an automated system is a separate issue. There is certainly more to go wrong on an ATO system. It is also a fact that the process of automation does not in itself improve the reliability of the electronic, electrical or mechanical equipment - these remain as issues that must be addressed separately
(c) The matter of ATO reliability when interacting with passengers on a high-density service certainly needs more thought. Computer controlled machines tend not to react well with humans (consider a room full of senior citizens and a video recorder).
2. Will it allow for faster journey times?
While automation could result in higher speeds, there is a finite limit to what can be achieved in existing tunnels. In theory you could bank the track round a curve, but you cannot do much with a dip. The fastest speed is not determined by the performance of the train, it is governed by what may be operated safely and comfortably – this also applies to braking. It is true that ATO trains have been allowed to drive at speeds that manual operators would never have been permitted to, but it ended in tears when the trains tried to show off. Now the computers have had the programming equivalent of a slap on the wrist and their speeds have been reduced. There may therefore be some extra speed, but it will probably not be as much as had been hoped for, at least not in the immediate future.
Also, as ATO goes system wide there is talk of ‘simplifying routes’. If implemented (and this seems to be a plausible option), it will result in more passenger changes and therefore a greater number of people in the system at any time. Route simplification would also introduce the secondary (though temporary) problem of disruption caused by the essential redevelopment of interchanges.
The bottom line is; the trains might travel faster, but the passengers might not.
3. Will we get higher line capacity?
Remember the maxim, ‘a motorway is the quickest route between two traffic jams’; you may gain a little extra time between stations, but I cannot see too many extra trains being packed in. We must still consider safety, so if you add more trains and reduce the distances between them you may have to reduce the line speed (see 2).
The argument that the new rolling stock offers greater capacity is not relevant as its design is not reliant on ATO.
Before we move on, let us put these last two points together. Simplified routes = more changes = longer journey times for passengers = more people in the system for longer = more congestion. Yup, we are going to need all the extra capacity we can get.
4. Will it be safer?
Well, you do eradicate the possibility of much human error, but you will not rule out the possibility of a programming error, of an equipment breakdown or of a type of leaf on the track that is not recognised by an onboard computer.
As newly constructed lines, DLR and the Victoria Line were able to run a real-time. experimentally service for an extended period prior to opening. The Central did not have that luxury and it showed.
To sum up: The changes that will occur as a result of the introduction of ATO may be noticeable in ways that were not really intended. While system wide ATO will be expected to enhance the service, there will be instances where it will not. That is a fact that the designers of the system will be well aware of; and it does not need a computer to work it out. As in any pioneering system, there will be several steps forward, but do not be surprised to see a few steps backwards as well.
