Penn Central RR computer system failure 2147

The computer systems were essentially efficient databases that kept track of car locations and waybill information, like who the shipper was, the car's weight, and where the car was destined. The car locations could be identified as a specific customer track, a yard in general, or on a particular train.

When the companies merged, they were unprepared for the changes, not only in the computer systems, but in the operations. Everything happening at once made for a complete mess on the first day, and it never really was cleaned up before the failure.

As an example, prior to the PC, railroads coming into Buffalo from the west could interchange to either the NYC or the PRR. Appropriate transfers would be operated to each railroad's yard with the cars destined to that railroad. After the merger, the NYC and PRR essentially became one, and the connecting railroads simply dropped off all cars at the most convenient yard, no longer separating them. This meant that the PC not only had to sort out the cars, something they previously didn't have to do, but also had to operate their own transfers between the yards to get cars where they needed to be for further handling. All this added to workload, rather than reducing it, and delayed shipments.

The computer systems essentially had similar problems. The practice at the time was for a clerk to punch cards for all cars that arrived at a yard in interchange. The card would physically be put into a pigeon hole in the yard office that corresponded to the track the car was sitting on. As a train was buttembled in a particular track in preparation for departure, the buttociated punched cards and waybills would collect, in the proper order, in the pigeon hole for that track. When the train was complete, the deck would be removed from the pigeon hole and placed in a card reader to feed the information into the computer system. The information would then be printed and a train journal handed to the train crew, plus it would be entered into the computer system such that each yard along the way could pull off the information they needed about arriving trains.

As I understand it, the intention was to operate the two systems in parallel when the PC started, and gradually shift over to a single system as time permitted. A temporary interface was provided to pbutt data between the two systems, but it turned out to be inadequate given the amount of data flow. In some locations they provided terminals for both systems, and clerks had to enter data twice to keep track of trains moving between the two predecessor railroads.

The parallel operation was needed because of the sheer logistics in training a large number of clerks in a new system, plus the need to physically install a whole new communication and terminal network to replace the system being displaced. Everything was to be completed yard by yard as resources permitted.

In looking at the Buffalo situation as an example, when the connecting railroads dumped cars in one yard, the clerks in that yard had the task of entering the data, but could not readily transmit the information to their sister yard when cars were transferred. The data was therefore entered a second time by another clerk when the car arrived at the second yard. Cars were lost, delays accumulated, and yards filled up with cars not being moved.