UK industrial valve sourcing, automation packages and practical application support

ALLWAY VALVESIndustrial valve supply & selection Request a Quote

Technical article

Control Valve Hunting and Instability: Root Causes and Systematic Diagnosis

  • Control valve hunting — the persistent oscillation of a valve position around a setpoint — is the most common complaint about control valve performance in process plants, and in the majority of cases the root cause is mechanical rather than control system: deadband, stiction, and actuator-process interaction cause oscillation patterns that no amount of controller retuning can fully eliminate.
  • Stiction — the threshold friction force that must be overcome before a valve begins to move from rest — creates a characteristic “stick-slip” motion pattern that is highly destructive to valve seating surfaces, packing, and actuator components, and is almost always caused by over-compressed or incorrect packing rather than any inherent valve defect.
  • Distinguishing between valve-induced hunting, process-induced oscillation, and controller-induced instability requires specific diagnostic tests — including step tests, relay feedback tests, and valve signature analysis — that must be performed systematically rather than by intuitive guessing.

Understanding Hunting: The Three Root Causes

Control valve hunting refers to persistent oscillation in valve position — the valve continuously moves back and forth around its commanded setpoint without settling. This oscillation is transmitted through the valve to the controlled process variable (flow, pressure, or level), causing process instability that affects product quality, energy consumption, and safety system performance. Three distinct root causes produce hunting, each with a different signature pattern and a different engineering fix.

Deadband-induced oscillation (limit cycling) occurs when the controller integral action continuously winds up against a valve that is not responding to small output changes due to packing friction or positioner deadband. The controller output increases until the valve finally moves (when the accumulated output change exceeds the deadband), the process variable overshoots, the controller reverses, the output decreases, and the cycle repeats. The characteristic signature is a square wave in controller output and a triangular wave in process variable, with an oscillation period equal to approximately 2 × deadband / (controller gain × process gain).

Stiction-induced oscillation (stick-slip) produces a different pattern: the valve sticks at a position, the controller output continues to change as the controller accumulates error, the valve suddenly slips by a large amount when the accumulated force exceeds the static friction, and the resulting large flow change causes a large process variable excursion. The stick-slip pattern has a characteristic irregular, asymmetric appearance — large jumps alternating with periods of no movement — distinct from the regular square wave of deadband-induced cycling.

Gain-scheduled instability occurs when the control loop becomes unstable at a specific operating point because the combined process-valve gain exceeds the stability boundary at that point. Unlike friction-induced hunting (which can occur at any operating point), gain-induced instability is position-specific — the loop is stable at most valve positions but oscillates when the valve is at the specific position where gain peaks. This is most common in butterfly valves near the closed position (where the flow characteristic is very steep) and in liquid level loops where the process gain changes dramatically with tank level.

Diagnosing Stiction: The Sticky Valve Problem

Stiction is the static friction that must be overcome to initiate valve movement from rest — distinct from kinetic friction (the friction during movement), which is always lower. A valve with high stiction requires a large initial force to begin moving, then moves with the lower kinetic friction until it stops and stiction must be overcome again for the next movement increment. This stick-slip behaviour creates jerky, irregular valve movement that is poorly controlled by any standard PI controller.

The stiction test for a control valve uses a series of small increasing steps in controller output from a starting position and observes when the valve first moves. If the valve does not move until the output has changed by 3% (for example), then moves by 4% in a single jump, the stiction is approximately 3–4% of span. A valve with stiction below 1% of span is considered good; 1–3% is marginal; above 3% requires maintenance intervention — typically packing adjustment or replacement.

Over-compressed packing is the most common cause of high stiction. Maintenance technicians who tighten packing to prevent leakage without measuring the resulting friction may compress the packing far beyond what is needed for sealing, dramatically increasing the stem friction. Live-loaded packing systems that maintain a calibrated compression load solve this problem — the load is set to the minimum required for the specified sealing class, rather than tightened to whatever feels tight enough. Maintenance procedures should specify the torque or extension for gland follower nuts based on the packing system design, not left to operator judgment.

The Relay Feedback Test for Loop Characterisation

The relay feedback test (also called the Åström-Hägglund test or ultimate cycling test) is a safe, standardised procedure for characterising the dynamics of a control loop from the closed-loop response to a specific input disturbance. It involves replacing the PID controller with a relay (on/off) element that switches the controller output between two values based on the sign of the error, inducing a sustained oscillation. From the oscillation period and amplitude, the ultimate gain and ultimate period of the loop are estimated and used to calculate PID tuning parameters.

The relay feedback test is valuable for diagnosing valve-induced hunting because it induces a controlled oscillation that reveals the loop’s dynamic characteristics. If the valve is hunting due to high stiction, the relay test produces an irregular oscillation with stick-slip features rather than the smooth sinusoidal approximation expected from a linear system — this directly diagnoses the stiction problem. If the valve responds smoothly to the relay and the oscillation is regular, the hunting is likely caused by controller tuning rather than valve mechanics, redirecting the diagnostic effort appropriately.

Remediation Strategies for Different Hunting Types

Once the root cause of hunting has been identified, the remediation strategy follows directly. For deadband-induced limit cycling: reduce the valve deadband by upgrading the positioner to a digital design (reducing positioner deadband from 1–3% to 0.05–0.2%), and review packing compression to ensure it is not unnecessarily adding to deadband. After valve mechanical improvement, retune the controller with tighter parameters to take advantage of the improved valve performance.

For stiction: reduce the packing compression to the minimum consistent with the required sealing class, using a torque wrench to the specified gland nut torque rather than tightening by feel. If the minimum compliant packing compression still produces unacceptable stiction, upgrade to a lower-friction packing formulation (PTFE-based rather than graphite for ambient-temperature service, or TFE-impregnated graphite for high-temperature service where pure PTFE is not suitable). If stiction persists after packing optimisation, inspect the stem for corrosion, scoring, or dimensional deviation that may be creating a mechanical interference with the packing bore.