16.1, PHASE 1, CAR RECORDS AND CAR TRACING
It is now time to leave the personal side
of life to write up more about the major Traffic Reporting and Control
System for which I had been selected as Project Manager. Once staff had
been selected, office space created and Southern Pacific software procured
as outlined in Chapter 14 it was necessary to consider the implementation
and further refinement of this immense project which eventually cost $70
million in cash and 700 man-years of effort. Charley Armstrong had originally
assumed that the S.P. T0PS System could be implemented in one package on
C.N., but more detailed study by our computer experts, of which Al Pozniak
most stands out in my memory, clearly showed the project with its needs
for some revisions to the SP software, (such as including trailer and container
records), plus the provision of some 300,000 circuit miles of communications
to serve over 2000 field terminals and the training of about 11,000
field and office staff to utilize the new system, would need to be designed
and implemented in stages which we called Phases. Also, those parts of
the phases requiring field startup would have to be done in geographical
steps across the railway.
Hence, for Phase 1 it was necessary to introduce
the transmission on data respecting car movements from the field yard offices
to the central computer files. This phase made it possible for anyone interested
in a car location to put a simple enquiry into their computer terminal
and within a few seconds get a message showing its last reported location
and status whether in a yard, on a train, loaded or empty or on a repair
track. We started implementation in the East, the first terminal to be
done was Sydney Nova Scotia. Charlie Armstrong called out a headquarters
business car so he and I could be on the ground for startup. At midnight
on the chosen day it was necessary for the local yard staff to enter the
current data on all cars in the yard into the system. When it was completed
about 2.a.m. we were all amazed at how we could now get not only the data
on an individual car but also a complete printed list of all the cars in
the yard. The full import of what we were into only struck us when we saw
the real list spilling out of the yard office terminal. To me I could now
see that these car records could be on train consists as well and the old
"train journals" or "wheelage reports" which were written in my dear old
dad's copy-plate handwriting in Rivers Yard Office would henceforth be
produced on the machine, provided of course that the basic location and
data status keyboard entries were made in timely fashion. Moreoer, Phase
1 was psycologically important to the success of the project because it
produced a useable and simple output that was available to staff in all
departments, not only transportation, but also mechanical, sales and marketing.
Customers could henceforth be given the latest information on their cars
in just a few seconds while they held the phone while our clerks
got the data from a simple machine enquiry! This had a very definite effect
in getting everyone to realize that the system was worth having and that
it could be done in manageable bites! If we had not been able to produce
something useable in Phase 1 it is quite possible the system would have
been derailed right there!
PHASE 1, COMPUTER HARDWARE AND COMMUNICATIONS
I will try in this section not to make it
too technical, mainly because I never was a real technical expert, but
more of a person who was responsible for directing the TRACS project from
a user and administrative point of view. Perhaps it's just as well as the
main frame technology on which the system was based is little mentioned
in the ebullient media to-day (Feb. 2000). To-day's computer thrust is
mainly in the field of the amazing proliferation of micro-processors
into everybody's life--there is probably one of them in 50% of our homes
in North America to-day, including the one on which this odyssey is being
typed! Main frames are still very important in the larger corporate and
government systems but one hardly ever seems to see or hear any reference
to them in the public media. The fact is that the nerds have taken over
and they are doing things for nerds and controlling the huge mass market
for computers which generates most of the revenue and hence media advertising
and interest in the microprocessor area.
However, TRACS, as a large on-line, real time
transaction oriented system handling over half a million transactions per
day and controlling the huge communication network had to be based on some
of the largest main frames available. Large amounts of on-line disk
memory was required to store the mountains of information as well as the
operating system programs. For example we had one huge file called UMLER
for Universal Machine Language Equipment Record which contained all the
physical data concerning all of our freight cars plus all of those on of
other railways because, as any of you who are train watchers would know,
the 2,000,000 car North American fleet circulates on all of our railways.
Thus we had two IBM 370 system main frames
initially based in Montreal and using IBM's standard operating system.
Programming language for the user programs was called Topstran and was
inherited from the Southern Pacific. It was basically a machine language
with some built-in utility programs specifically designed for railway needs.
It was slow work for our programmers, but it was very efficient in the
use of memory and other expensive computer resources available at that
time. For example, one megabyte of memory, which is much less than
you get in to-day's desk top $2000 microprocessors, cost hundreds of thousands
of dollars. Initially we ran the system on just 4 megabytes of memory--later
on we had to increase it to about 12 megabytes when we changed from the
DOS to the MVS operating system.
Because the glamour of computers so caught
the public eye from 1950 onwards the importance of communications in this
new world was often overlooked. Let me emphasize here that for large spread
out systems such as TRACS communications was as necessary and as vital
to the existence of these systams as were the computers. Consider first
that Southern Pacific's initial try at TOPS development in the early '60's
failed, chiefly because existing telegraph and telephone wire circuits
were unable to support the needs of the system. They were only successful
when they went to the use of Microwave radio. In the 1980's our efforts
to get TRACS installed in India failed partly because their pole line based
wire circuits were unreliable.
Fortunately, when we started implementation
on the CN in the 1970's we already had a trans-continental Microwave system
for all our long distance needs. It was partly a byeproduct of the needs
for capacity to transmit television pictures so CN/CP Telegraphs could
compete with Bell and the provincial phone companies for this type of business.
I may have mentioned earlier in this chronicle (1953-54) that Bell had
won the CBC television business away from CN/CPT at that time. At any rate
CN/CPT's system proved to be great and the microwave in fact had a lower
failure rate than the main frame computers! It only went down a few times
each year when the Aurora Borealis was active during extemely cold early
morning hours on the prairies. To ensure microwave reliability under normal
conditions there was always one circuit available for fallback. To connect
between our computers and CNT's communicatons centre in Montreal we had
a muti-circuit cable supplied by CNT plus one supplied by Bell for backup;
this one followed a different route under the streets so that if someone
accidentally dug up the CNT cable we would not have our whole system down!
16.3, PHASE 2 GIVES EMPTY CAR CONTROL
One of the great problems in railroading which
had bugged us from the earliest days was that of knowing when cars
were made empty and then directing them to the optimum point for re-loading.
Loads did not present a problem because every car was accompanied by a
waybill giving the shipper's instructions as to where the car was to go.
When railroads first started they were short lines only running from point
A to point B and back so car control was then not a problem for loads or
empties. However, as railroads grew and multiplied there were literally
millions of combinations of origins and destinations with cars moving over
long as well as short distances. Phase 2 of the TRACS system was developed
so as to take in the information reporting a car's unloading and matching
it against customers' car orders to find the optimum point to which to
send that car. One of the key features in addition to customer priority
was that the program was designed to search for the nearest point needing
that type of car, going in ever widening circles until a hit was made.
The effect of this was that empty car mileage was shortened because locally
available empties could be used where suitable. the results of this empty
car control system was that each car got to handle a new loaded cycle every
15 days instead of 19, and this meant the freight could be
moved with about 20% fewer cars! The result was quite visible in that the
C.N bought very few new cars in the 4 or 5 years following the implementation
of the system. This represented a capital saving of several hundred
million dollars in the car fleet and was the main identifiable financial
benefit of TRACS!
16.4, PHASE 3, MEETING MARKETING NEEDS
Once the basic operational needs were fulfilled
it became necessary to consider how the system could better serve the requirements
of Marketing and Sales. Modern trends in railway thinking driven in part
by highway trucking competition had started to place greater emphasis on
these departments as distinct from operations alone. Thus the Marketing
people worked with us and we with them to find what we could do to help
them. This resulted from a broad study guided by marketing wherein they
developed 22 separate strategies to improve their activities. Not all of
these strategies were system related, but a number of them were. Empty
Car Management, for example was number three on their list enabling them
to make suggestions whereby that system could give better customer satisfaction
re empty car availability. Encouraging direct customer access to the system
for customers tracing their own cars was another. Concerning the improvement
of efficiency in originating customer loads was achieved through setting
up permanent waybill patterns in the local terminals for all regular movements
of five cars or more in a year. All the customer needed to do was to phone
in the number of the car he was loading and quote the number of the applicable
waybill pattern and the machine would do the rest. The system also facilitated
the notification of customers whose shipment might be delayed because the
car containing it was "bad ordered". Through meeting their needs we got
the support of the Marketing and Sales department without which the system
would not have received continuing support from top management for the
funds we required each year to build and operate the system.
16.5, FIELD IMPLEMENTATION AND TRAINING
Field implementation was mainly part of Phase
1, but it did carry on beyond that as various technical improvemets were
put in place and other functions which were not in the original Southern
Pacific TOPS system were added.
The main part of the Project was completed
between Feb. 1971 and July, 1975. Field implementation started on the Atlantic
Region in 1973 and finished up on the Mountain Region in Vancouver. At
some point the field implementation was considered top priority and I was
given the title of General Manager of Implementation. The continuing
work of programming was given to Al Pozniak together with programming jobs
for other systems.
The key field job was not that of installing
hardware nor providing communication but was the training of yard
office staff, yard engine crews and train crews in the use of the new system.
The yard office staff training was the most intensive because they were
the main ones who had the job of entering the data as well as making use
of it; the real complexity of railway operation is in the yards, where
trains are assembled or disassembled, not in the point to point movement
of trains on the main lines. To do the field training we selected a group
of 50 sesoned yard office employees from across the railway and gave them
a month of intensive trainig in TRACS procedures. We then divided them
into teams and sent them out to the field areas to train the people in
every yard. These teams were leap-frogged across the country as one yard
was completed and another started. We flew them from one point to another
and they became known as the "flying 50". One young fellow from the B.C.
area who was of Asian extraction, picked up the name of "China Clipper"!
The teams were put up in motels and they worked rotating hours with all
three shifts in the week when actual implementation was taking place. They
would stay to back up the regular yard staffs until they were sure the
regulars could handle it without further assistance.
For the pre-implementation training we had
a number of special truck trailers fitted out as classrooms complete with
data terminals. These vans could be handled to the next yard by piggy-back
rail and spotted near the yard office for the duration of the training
and implementation. When spotted they would be hooked up with a local drop
line to the CNT so they could actually work with the main computer system
which had a special training mode for the purpose.
To co-ordinate the implementation it was necessary
to set up a "war room" in our Montreal Headquarters which had speakerphones
connected to direct phonecircuits into the yard offices being implemented
on that cut. Implementation day was always on a Sunday when traffic was
light as the process was easier on a light traffic day where yard operations
had to be put on hold while all the local car records were entered into
the new system by the local yard staff. The main person in charge was the
Regional General Superintendent of Transportation as he was the voice of
authority to his own field people whereas they would not have known us
project people in the same way. We of course backed up the Gen. Supt. in
the war room with our senior implementation people and with technical programmers
to handle any program problems. In any case the whole arrangement worked
very well and in only one case across the country did we have an imperfect
flow of data and have to send some field implementation people back to
do some retraining.
The only other hitch we had did not delay
progress but is worth mentioning. We had found that implementation and
training went the slowest in the Southern Ontario District where traffic
was most dense and train runs between terminals were short. We were trying
to implement there before Christmas and the system seemed to get bogged
down at times. Doug Fletcher was then the Operations Manager on that area
so he and I agreed to keep plugging away with extra training help until
the normal Chriatmas-New Year's season traffic slump. It paid off as the
staff got hold of all the procedures firmly during the ten days of low
traffic and never looked back when traffic rose to normal levels after
NewYear!
However, that was not the end of it. The General
Manager of the Prairie Region based in Winnipeg heard of the problems in
Ontario and decided he would not allow the system to proceed west of Armstrong,
Ont., where the Prairie Region began. So a message to this effect was sent
to me through the medium of Johnny Clark who was the Prairie Region GST.
My first reaction was to go to Jack Spicer, who, as Executive Vice-Pres.
was my boss, and have him raise hell. Instead, after some thought I went
to Ross Walker, the Chief of Transportation, who had a fairly cool head.
We decide to ignore the message from Winnipeg and wait until the good news
that the SOD problems were solving themselves to percolate, as it would
be at least 6 weeks before we would have to go beyond Armstrong. The good
news did its work and when we moved our implementation team into Winnipeg
there was no opposition. Johhny Clark was free to give us all possible
help and he and his local people actually cooked up some interesting twists
to enhance the training. For example, they got us to liken the system to
a football type training diagram showing how the data had to flow from
the various players in yards, on trains, in depatching offices etc. which
made it like a game thry understood. The moral is, "Don't start a war in
high heaven if the problem can be solved by gentler means on earth!"
16.6, COAST TO COAST, TRACS ENDS IN VANCOUVER
So in July 1975, after almost five years with
a caste approaching 250 people at its height, we were able to implement
the B.C. area. We felt it warranted a bit of a celebration so we arran
ged for a couple of business cars to be spotted at the passenger
station and issued invitations to the media to come for lunch with the
Regional Vice President, who by this time was Charlie Armstrong. We set
up a couple of terminals in the cars and got some photogenic lady yard
office staff as operators so that we could demonstrate car tracing, train
journals, train times etc. It was the first time I had ever done anything
that was worthy of press attention and I was proud of it. However, pride
goeth before a fall and we only got one junior reporter--all the other
invitees had gone out on a tour of Okahalla prison from where a dangerous
murderer had escaped. While we did get a picture of one of the terminal
operators in the newspaper there was no covering story to represent the
immensity of the project and its importance to the railway and commercial
world, just completed right here in Vancouver! In any case, it did drive
home the media adage as a lesson to us all, "if it bleeds, it leads."
Anyway, we did have a great followup party
that evening in the Vancouver Hotel for the benefit of our local and visiting
team members. We had lots of merriment and I recall at one point the waiters
carrying candles led us in a snake dance around the room! The other benefit
for me personally was the opportunity to attend my son Jim's graduation
from Simon Fraser University where he had completed his studies in Computer
Science and Education! How he got that far is a story to be told in a later
chapter where we will bring all our family events up to date.
16.7, SOME IMPORTANT ADDITIONS TO TRACS
We found out that a system as large as TRACS
takes on a life of its own and requires maintenance and the ability to
be kept up to date with technical or operational changes needed by its
user departments. One of the early changes we introduced was to shift away
from what had been the traditional punch card system for handling data
to the use of intelligent input/output terminals which had some memory
capability of their own. The terminal we selected was the Datapoint 2200
which served us well and permitted the elimination of card racks in each
yard where each car in the yard had a card which had to be moved from one
hole to another in the rack to reflect physical movement in the yard. It
also permitted us to set up things like the waybill patterns mentioned
earlier, as these could be stored locally and then called up on the datapoint
screen to fill the blanks after which the data could be transmitted to
the main system.
The TOPS system we got from SP did not allow
for the handling of trailers and containers once they were off the flat
car so we added to TRACS the capability to enter trailer events right to
their final destination by road. With the continuing rise in intermodal
traffic this added capability was quite important to the railway and its
customers. In larger yards we found it advantageous to develop a yard sub-system
running on its own mini-computer which would keep up to date on the yard
switching movements in detail. Only when the car was placed on an outbound
train would a report be sent to the main central system. The software for
this internal system was developed by Bob ?
who was in charge of CN's Research Department. They demonstrated
it on a pilot model and we took it from there after a successful field
trial.
After operating the system for a time based
on a backup computer in Montreal (which was used for batch processing which
could be set aside if the machine was needed for TRACS) we decided that
it was not secure enough to have both machines in the same computer complex.
Our alarm was heightened by the event wherein a group of disaffected students
trashed the Sir George Williams University computer room, even throwing
some components into the street. After a survey and an examination of the
data flows across the railway we found that the traffic centre of gravity
was moving westward and the most favorable place to have a backup location
was Winnipeg. So a new special computer building was set up off Osborne
St. near one end of Fort Rouge Yard. As the time came to commission it
we realized that having to have round the clock staff there and not much
access to batch jobs as filler it made sense to run the main on-line Tracs
system in Winnipeg with the system in Montreal to be the backup. We also
enhanced Montreal's position by installing a quick-starting 1000 horsepower
gas turbine to provide power in case of electric power system failures.
I got a bit of a personal glow from realizing that as a native of Manitoba
and a graduate of the University of Manitoba in Engineering I had been
instrumental in establishing a key front line high-tech development in
my native land!
There were over the years many more changes
and additions to TRACS which I will not report here. They will possibly
show up in subsequent chapters having career work references. Following
the successful completion of TRACS I was given the task of planning what
the future might hold for C.N. in meeting the growing computer system needs
for all departments. I had hoped I might return to the Transportation department,
but the open position of Asst. Chief for Car Service was denied to
me. However, they offered me the position of Chief of Management Services,
so with every Engineer's hope to become Chief of something I accepted.
It was actually quite an important job embracing Operational Research,
Industrial Engineering, and Office Building maintenance as well as all
the major computer systems.
As for TRACS, it remained in its basic form
for 20 years until it was supplanted the by a more modern software technology
developed by the Santa Fe railway. This system is called SRS and has the
important capability of being able to predict the schedules likely to be
followed by each car and to make that information instantly available to
customers as well as to railway staff! I can still take satisfaction from
the fact that TRACS lasted as long as it did, as the life of any computer
system in the last half of the 20th century was lucky to exceed 10 years.
And now dear reader, we are finished with
the story of TRACS, which was easily the greatest of my working accomplishments
in life. Now it is time to step back and fill in the equally important
events of personal and family life which I have neglected to record in
passing through the five years of TRACS!