Month: December 2018

I went from college into a career with DuPont starting in 1976. After an initial assignment, I worked in a process that used a DEC PDP-8 computer for data monitoring and for control of certain critical process parameters. This was in a process that produced hydrogen cyanide, so reliability of the computer system was critical. This computer, when it was booted up, required setting toggle switches in order to start the sequence. Then a paper tape was run through a reader, and the system would lurch into operation. This modern machine also used punch cards for program input.

By 1984, our company began to use IBM PC’s. To have this type of power upon your desk was amazing. These were not used for process control, but enabled us to have the power to write and distribute through an e-mail system documents that bypassed the old strictures of communication. If you can imagine now living within a hierarchical system that required all communications to be written by hand, approved by supervisors, typed by a secretary, then copied and sent through corporate mail systems, that was the world as it existed in my company. It was the same as existed in most other companies around the world. It was also a first introduction to the ability of technology to replace jobs. Secretarial positions shrank in number once they were not needed to serve as a key link in the communication process. Those who remained either had to be flexible enough to pick up other skills, or became administrative assistants to those high-ranking administrators where it was still valued to have someone to serve as an intermediary.

In the late-1980’s, the chemical process I worked with had a computer used exclusively for process monitoring. We had gotten past the toggle-switch and paper tape process, but I learned techniques for data compression. For each variable that was monitored, you got to choose how much change you would allow in the value before another data point was recorded. Computer memory was still limited, so it was necessary to use a bit of judgment to tweak each setting so that any signal noise was eliminated, but significant changes in variables were recorded and could be graphed. This computer also held the statistical program Minitab, which helped in determining correlations and other relationships between variables. I began using that program in 1991 to start tracking the performance of my 401K investments, a spreadsheet I maintained until my retirement in 2014.

This process had computers to monitor variables, but it still had individual control loops. Each variable that needed to be controlled had a piece of equipment on a panel board. Out in the field, a sensor would provide a reading that would be transduced into a 3-15 PSIG (Pounds per Square Inch Gauge pressure) signal in the field. That signal would be fed into a small metal tube and routed back to the control room, where it got transduced back into an electrical signal, and fed into the controller. We would use controller logic to provide the optimum settings for this particular loop to keep the loop stable. There was also a signal splitter that sent the signal to a chart recorder, where a paper chart was fed through and multi-colored inks were used to display multiple variables onto a single recorder. Normally there was a maximum of three variables on a single chart recorder. The electronic signal was also sent to the monitoring computer as well. Now, if modifications were made in the field, say for a new piece of equipment, it would require running a new piece of tubing from the field back to the control room, placing a new controller into the metal board, and installing all of the sensors and transducers to enable the system to work. The entire process was labor and capital intensive, and required a significant amount of operators, electrical and instrument (E&I) mechanics, and engineers in order to maintain a plant and keep it operating safely.

During the late 1980’s and early 1990’s, the next change in process control occurred. Distributed Control Systems were commercialized. These computer systems replaced all of the controllers on a panel board with two computer consoles and a pair of keyboards for entry of commands. These computers used wire pairs directly from the field to provide their input, so no longer were 3-15 PSIG transducers or metal signal tubes needed. You always installed extra wire pairs in a wire bundle from a field signal box back to the DCS so if you expanded the number of controls or signals in the field, it only required installing the last leg of the wiring in the field.

The displacement of workers with computerization was huge with this step. The number of E&I mechanics used to keep these systems up was much fewer than before with all of the signal transducers and individual controllers and chart recorders. Chart recorders were totally dispensed with, as all records were retrievable via computer. And fewer control room operators were needed, since no longer was it necessary to go up and down the control panel and record readings every few hours. A single operator could maintain the entire process by himself, and the back-up operator could be assigned into the field for a portion of his shift (most but not all operators were male). Even with engineers, there were fewer needed, since control loop tuning was all but eliminated with the new algorithms available in the computerized systems. There is no wonder why the population of workers in my plant kept going down, year over year. It became a ritual that every 2-3 years, we would undergo a purge of excess people. Not all of it was due to automation, since world economic conditions rendered multiple chemical processes uneconomical, but at least half of the reductions in force were due to automation.

So during the roughly 15 years while I was directly supporting chemical manufacturing, constant changes in technology kept paring the need for employees of the company. At the same time, the support staff kept shrinking as well. Whereas we once had an entire group of workers tasked with maintaining and updating blueprints (I can still remember the ammonia aroma of a freshly printed blue-print), they left once all print updates were done on the computer. Since most documents traveled by e-mail, the need for physical mail distributors went way down as well. Combine that with the growing international competition in the chemical industry, and you will understand why the plant I worked at in West Virginia did not hire any hourly employees for a period of 20 years. If you really want to know why the middle class has atrophied in the US, just look at the jobs that were displaced due to technology improvements during the time from 1975-1995. And the technological changes have only increased since then. That is why the talk about Making America Great Again by revitalizing manufacturing rings hollow. The direction manufacturing has taken involves replacement of people by technology, allowing a smaller number of people to maintain a growing production output. We’d best be thinking about how to restructure the workforce to pay wages that reflect the value society places upon the work, rather than weigh everything on the scales of economic efficiency.