A single pump provided a hydrocarbon feed to a manifold serving four parallel reactors, each with its individual vaporizer. Each vaporizer was a smallish vessel with some heat exchange coils to thoroughly vaporize the stream for the ensuing reactor. The process superintendent was unhappy with the constant cycling he observed, not only in the pressure and flow from the pump, but also in the vaporizer levels. If the level went high enough in any of the vessels, liquid feed would burp over into the outlet and the excess flammables would likely shut down that train. If the level went too low and starved the reaction, a shutdown would also be invoked. The cycling had gone on for some time, and one might conclude from its persistence that operations was accustomed to it.
While the plant had updated to a modern, microprocessor-based control system from a major DCS vendor, attempts to improve the level control tuning were always unsatisfactory—the cycling went on.
Creative and thoughtful systems engineers added feed-forward for vaporized flow rates, not unlike steam drum controls. They were disabled because, if they had any impact at all, they just made matters worse. Tuning seemed impossible. Tweaking gain and integral action can do little to attenuate a cycle if the loop in question (the vaporizer level, for example) wasn’t the loop with a problem. What loop, then, could it be?
If the system had been constructed in, say, 1963, chances are all controls would be pneumatic. And level control, in particular, would likely be “local”—integrated with the displacer level measurement device. Clever and inscrutable to the uninitiated, rugged motion-balance mechanisms, torque tubes, flapper-nozzles, bellows and bourdon tubes yielded proportional control for a control valve and in most cases, reliable level control.
One item that lingered on in the modernized control system was the pneumatic valve positioner. In the 1980s and 90s, as we hastened to fit our plants with emerging DCSs, often the pneumatic field devices—transmitters and positioners alike—lingered on, connecting to the new electronic I/O through massive banks of P/I (pneumatic to current) and I/P converters. At best, the field devices might be upgraded to an electronic version, but valve positioners still remained pneumatic, modulated by a field mounted I/P converter. So it was with the vaporizer level controls—I/P converters sent a pneumatic signal to each control valve’s pneumatic positioner.
There remains a loop that has been in the field since the middle of the last century (at least) and remains there today. System engineers might dream of elegant control systems, but none of it works if the innermost loops fail to function. What couldn’t be easily seen from the comfort of the engineer’s office was the significant hysteresis in the valve position “control loop.” Worn or loosening linkages and cams meant the valve position wasn’t consistently moving when the signal from the house was ramping up or down, which wasn’t obvious unless your suspicions led you to the physical device itself.
The “innermost” loop of 'most of our control schemes has long been “in the field” and remains there today, but there are many potential improvements if our plant has the infrastructure to support them. No longer a proportional-only, motion-balanced, flapper-nozzled and mechanical engineer’s amorous dream, today’s positioner sports tunable three-mode control, onboard diagnostics, trending, and (in the fieldbus version) a repertoire of function blocks including PID. Valve position versus setpoint can be trended "in the house" for those with fieldbus or HART-capable systems. Still essential for even rudimentary tuning to be effective, why not retain its role as an effective platform for truly distributed control?
Clearly the power and affordability of capable chip-ware and memory means we can still contemplate the inherent advantages of control loops closely coupled to their final control elements. Clouds of I/O divorced from any controls appliances may be our future, but perhaps “control anywhere” should continue to be accommodated for the field devices that have excelled at it.