Separating, Filtering, Clarifying, Precipitating, and Still Machine Setters, Operators, and Tenders

• Batch stills1853–1900

  Batch stills, manual filter presses, and gravity separators

  The original tools of this trade were batch devices: a petroleum still was a sealed iron vessel heated from below by a wood or coal fire, with the operator ("still man") controlling temperature by regulating the firebox, watching the color and viscosity of the distillate running from a condensing coil, and deciding by sight and smell when to switch cuts between fractions. Filter presses required the operator to manually load slurry-filled filter cloths between the press plates, tighten a large central screw until the desired pressure was reached, and wait while liquid clarified through. Gravity cream separators in dairies required the operator to pour whole milk into a shallow pan and wait hours for cream to rise before skimming. These were physical, sensory, high-judgment operations with no instrumentation beyond a thermometer and a clock.

• Continuous centrifuges1877–1940

  Continuous centrifuges, automated filter presses, and continuous distillation columns

  Gustaf de Laval's steam-powered centrifuge (1877, patented with Oskar Lamm in 1878) replaced hand-cranked and gravity separators with a machine that ran at 4,000 rpm continuously, processing milk or chemical slurries far faster than any batch device. Commercial centrifuge manufacturing expanded rapidly after de Laval's introduction, with dozens of manufacturers entering the market by the 1890s-1900s. In petroleum refining, continuous distillation (column stills, replacing batch pot stills by the 1880s-1900s) fundamentally changed the still operator's job: instead of managing a batch cycle from firing to quenching, the operator now maintained a steady-state column, monitoring tray temperatures and side-draw compositions to hold the target cut points. The operator no longer judged "when is this batch done?" but "is this column in equilibrium?" A new skill set emerged: reading pressure gauges and temperature indicators, adjusting reflux ratios and steam injection, and troubleshooting flooding or fouling. The Dehne filter press acquired mechanical screw drives and eventually hydraulic plate-closing mechanisms in the early 20th century, reducing the physical labor of tightening, though cloth-changing remained manual.

  Effect on the work

  Continuous centrifuges and column stills were more productive per worker than their batch predecessors, allowing total industry output to grow without proportional headcount growth. However, the sheer expansion of the petroleum, dairy, and sugar industries through this period meant total employment in these roles grew substantially even as productivity per operator rose.

• Panel boards1940–1975

  Panel boards, local instrumentation, and analog process control

  By the 1940s petroleum refineries and chemical plants had installed panel boards in local control rooms: rows of pressure gauges, temperature indicators, and flow meters, each wired to sensors on a specific piece of equipment. The operator's job became reading this panel, correlating readings across multiple instruments, and manually adjusting control valves (by turning handwheels or pushing levers connected to pneumatic actuators). Pneumatic control loops, using compressed air to operate control valves automatically against a set point, became standard in the 1940s-1950s and automated many routine adjustments. But the operator still needed to be present: set-point changes, equipment upsets, transitions between operating modes, and the inevitable instrument failures all required physical attention at the panel or at the equipment itself. Shell Oil's "Royal Dutch/Shell" refineries pioneered systematic operator training programs in the 1940s, formalizing what had previously been passed down informally from senior "still men." The job became more cognitive and less physical but remained intensely local: you had to be in the control room, next to the panel, to see what the plant was doing.

• Distributed Control Systems1975–2000

  Distributed Control Systems (DCS) and control-room consolidation

  Honeywell and Yokogawa introduced the first commercial DCS in the mid-1970s, replacing hardwired panel boards with digital controllers linked by a data highway. A single DCS workstation could display trend data from hundreds of sensors across an entire process unit, and operators could change set points or adjust control loops from a keyboard rather than walking to a local panel. Refineries and chemical plants began consolidating control rooms in the 1990s and early 2000s: where previously each process unit (a crude distillation unit, a naphtha reformer, a lube oil filter plant) had its own board operator, DCS enabled one operator to supervise two or three units simultaneously. The OCAW lost approximately 50% of its membership between 1980 and 1995, partly from refinery closures and partly from this headcount reduction per unit of capacity. The DCS also changed the quality of operator judgment required: with high-frequency data logging, trend displays, and automatic alarms, operators shifted from continuously monitoring gauges to responding to abnormal situations and optimizing performance within the safe operating envelope.

  Effect on the work

  Control-room consolidation driven by DCS reduced operator headcount per barrel of refining capacity substantially during the 1980s-1990s. The OCAW, which had represented approximately 210,000 workers in 1956, represented only 80,000 at its 1999 dissolution -- a loss of roughly 60% over 43 years, attributable to both plant closures and automation-driven productivity gains per operator.

• Advanced process control2000–2020

  Advanced process control (APC), online analyzers, and remote monitoring

  By the 2000s most large refineries and chemical plants had layered advanced process control (APC) software on top of their DCS: multivariable model-predictive controllers that could simultaneously optimize dozens of related control loops against an economic objective, pushing yields and throughput closer to technical limits than any human operator could manually achieve. Online analyzers (near-infrared spectroscopy, online gas chromatographs, in-line viscometers) replaced many of the periodic laboratory samples the operator had previously walked to the lab to submit. The operator's role continued to evolve toward oversight and exception-management: the APC ran the unit most of the time, the operator intervened when the controller limits were exceeded or when a piece of equipment failed. Membrane separation technology matured in this era for pharmaceutical and bioprocessing applications, creating a new sub-segment where operators managed pressurized membrane systems for nanofiltration and ultrafiltration. Total employment in the occupation continued to decline gradually as APC and online analytics further reduced the human judgment required in normal operation.

• AI-assisted process optimization2020–present

  AI-assisted process optimization, predictive analytics, and remote operations centers

  The 2020s have introduced machine-learning-based process optimization tools that monitor sensor streams across entire facilities and suggest or automatically implement control adjustments to minimize energy use, maximize yield, and predict equipment fouling or failures before they occur. Remote operations centers now allow one operator to supervise multiple geographically dispersed facilities from a single location -- a model pioneered in oil production but increasingly applied to food processing, water treatment, and chemical manufacturing. For 51-9012 workers, this era represents a continuing shift from in-person process monitoring toward exception-management and system oversight: the routine tasks of reading gauges, adjusting set points, and collecting samples are increasingly automated or sensor-driven, while the remaining human value lies in diagnosing complex upsets, managing equipment transitions, ensuring regulatory compliance (FDA, EPA, OSHA documentation is a growing share of the role), and applying tacit process knowledge that has not yet been encoded in the AI systems. The occupation is not likely to be eliminated by this wave of automation, but the ratio of oversight to hands-on operation continues to shift.

What's changing in your day

Three parts of your work where AI is already doing real lifting, and what stays yours.