An oscillation is considered fast if it is less than twice the ultimate period. The ultimate period can be considered to be about 4 times the total loop dead time. An exception is the severely dead time dominant process where the ultimate period is about twice the dead time. Some sources of fast oscillations and examples of causes:
In all cases, it is assumed the subject loop is stable.
(1) Pressure fluctuations (e.g. oversized pressure regulation)
(2) Bubble formation (e.g. flashing and heat increase to boiling mixture)
(3) Bubble collapse (e.g. cavitation and cold liquid increase to boiling mixture)
(4) Dissolved gases (e.g. spargers and reactions)
(5) Poor mixing (e.g. insufficient back mixing - axial agitation)
(6) Sloshing and vortexing (e.g. vessel feed entry and exit)
(7) Flame instability (e.g. oversized burner and pressure fluctuations)
(8) Hammer (e.g. sudden large movement of liquid valves)
(9) Surge (e.g. sudden closure of gas valves)
(10) Interactions (e.g. poor tuning of fast loops that affect subject loop)
(11) Valve oscillations (e.g. poor valve positioner-booster tuning and too small an actuator size)
(12) Burst of oscillations (e.g. slow secondary loop)
(13) Decaying oscillations with period = ultimate period (e.g. subject loop PID gain too large)
(14) Decaying oscillations with period = 0.6 x ultimate period (e.g. subject loop PID rate time too large)
(15) Decaying oscillations with period = 1.6 x ultimate period (e.g. subject loop PID reset time too small)
The first 10 sources persist when the subject loop is put in manual.
If the oscillations are close to the ultimate period of the subject loop, the oscillations will grow from resonance when the subject loop is in automatic. If the oscillations are less than the subject loop dead time, resonance should not be an issue but amplification and perpetuation can occur. A high controller gain or rate time setting makes resonance, amplification, and perpetuation worse. A signal filter set just large enough to keep PID output fluctuations within the valve deadband and resolution limit may help. In general signal filters should be minimized because they add dead time to the loop or even worse hide process excursions if they become the largest time constant in the loop. If the filter is large enough to become a secondary time constant, the performance of integrating and runaway processes are seriously degraded.
The correction for sources 1-7 involves improving equipment, control valve, and pressure regulator design and installation. Keeping the liquid pressure above the vapor pressure at the sensors is critical. The correction for sources 8 and 9 involve slowing down the closure of the offending valves taking into account the installed flow characteristic of these valves and improving pressure and surge control. The correction for source 10 is to slow down the tuning of the least important loop or add decoupling or go to model predictive control.
The correction for source 11 is to tune the positioner for a stable and smooth response with minimal overshoot for actuator and valve combination. If a booster is used on the positioner output, a bypass valve around the booster must be opened to prevent instability (e.g. very fast limit cycle). If fast oscillations occur near the closed position, make sure the actuator is sized to handle at least 150% of the anticipated largest pressure drop and largest friction stem and seating-sealing friction. The correction for solids causing an alternating loss and burst of flow is a valve design where solids cannot accumulate. The correction for source 12 is a faster secondary loop and external-reset feedback to prevent the primary loop output from changing faster than the secondary loop can respond. The correction for sources 13 through 15 is less proportional, integral, and derivative action in the subject loop.
