Integral control eliminates steady-state offset by continuing to adjust the manipulated variable until the accumulated error is zero. In a liquid-level pilot plant, a proportional-only controller needs a persistent error to maintain the valve position required after a load disturbance. Integral action accumulates that error and changes the valve signal until the level returns exactly to its setpoint.

The essential principle is that proportional action responds to present error, while integral action responds to the history of error. Because the integral term keeps changing the controller output whenever error remains, the process can reach a new force or flow balance with zero steady-state offset.

Why Proportional Control Leaves Offset

The controller needs error to sustain a correction

For proportional control,

[ u(t)=K_c e(t) ]

where (u(t)) is the controller output, (K_c) is proportional gain, and (e(t)) is the setpoint error.

After a disturbance, the valve may need to remain at a new position to restore the material balance. A proportional controller can produce that changed output only if a nonzero error remains.

The liquid level settles at the wrong value

Suppose the outlet flow increases suddenly. The inlet valve must open further to balance the new outlet flow.

If the level returned fully to the setpoint, the error would become zero. The proportional controller would then return its output to its original value, the inlet flow would become too small, and the level would fall again.

The stable outcome is therefore a nonzero level error, known as steady-state offset. For a unit load disturbance in the reference model, the offset is proportional to (1/K_c).

Increasing gain reduces but does not remove offset

A larger (K_c) produces a larger corrective action for the same error, so it reduces the offset.

However, proportional control cannot make the offset exactly zero at finite gain. Excessive gain can also cause oscillation, poor damping, or instability.

How Integral Action Removes the Offset

Integral control accumulates error over time

Integral control is commonly represented as

[ G_c(s)=\frac{K_c}{\tau_I s} ]

where (\tau_I) is the integral or reset time.

In the time domain, the integral contribution is proportional to

[ u_I(t)=\frac{K_c}{\tau_I}\int_0^t e(\lambda),d\lambda ]

The controller therefore responds not only to the current error but also to how long the error has existed.

A small error continues to change the valve signal

Even when the level error becomes small, the integral term continues accumulating it.

That ongoing adjustment gradually moves the valve until the manipulated inlet flow exactly compensates for the disturbed outlet flow. At that point, no error is required to sustain the new valve position.

Zero error is the only steady state

For a finite, non-saturating integral term to remain constant at steady state, its input must be zero:

[ e_{\text{ss}}=0 ]

Thus, when the loop is stable and the actuator has sufficient authority, PI control drives the process variable to the setpoint:

[ y(t\rightarrow\infty)=r ]

For a unit setpoint step normalized to one, this becomes

[ y(t\rightarrow\infty)=1 ]

The steady-state offset is therefore eliminated.

What Is Special About Liquid-Level Processes?

The tank itself may be an integrating process

For a tank with constant outlet flow and inlet flow as the manipulated variable, the level balance can be written as

[ A\frac{dh}{dt}=q_{\text{in}}-q_{\text{out}} ]

where (A) is the tank cross-sectional area and (h) is liquid level.

In transfer-function form, the idealized process is

[ G_p(s)=\frac{1}{As} ]

The (1/s) term means that a persistent flow imbalance causes the level to continue changing rather than settle naturally.

Proportional control can track a level setpoint without offset

For an ideal integrating level process, proportional control can produce zero steady-state offset following a setpoint step. The level stops changing only when the flow imbalance has been removed, which occurs at the target level.

This is an important qualification: the claim that proportional control always leaves setpoint offset is not correct for pure integrating processes.

Load disturbances still produce proportional offset

A step change in inlet or outlet flow is different. Under proportional-only control, the controller must maintain a changed valve signal, and that requires a persistent error.

For a unit load disturbance, the residual offset in the reference model is

[ e_{\text{ss}}=-\frac{1}{K_c} ]

The sign depends on the disturbance and feedback convention, but the key result is that the magnitude is inversely related to proportional gain.

Integral action removes this disturbance offset by creating the required steady controller output independently of a persistent error.

The Closed-Loop Interpretation

Integral action adds a pole at the origin

The integral controller contains the factor (1/s), which adds an integrating mode to the controller-process loop.

This increases the system type and gives the loop the ability to reject constant disturbances with zero steady-state error, provided the overall closed loop remains stable.

The final-value theorem confirms the result

For a stable closed-loop system, the final-value theorem gives

[ \lim_{t\to\infty}y(t)

\lim_{s\to 0}sY(s) ]

Because integral action increases the low-frequency loop gain without bound in the ideal model, the steady-state error to a step reference or constant load can approach zero.

The theorem applies only when the closed loop is stable and the relevant final value exists. It does not guarantee acceptable transient behavior.

Understanding the Trade-offs

Integral action can make the response sluggish

A long reset time produces weak integral action. Offset is removed more slowly, and the process may take longer to recover from a disturbance.

A short reset time produces aggressive correction, which can improve recovery speed but reduce damping.

Poor tuning can cause overshoot and oscillation

Integral action increases the dynamic order of the closed-loop system. If (K_c) is too high or (\tau_I) is too small, the level may overshoot the setpoint or oscillate around it.

The correct objective is not merely zero offset. The loop must also have acceptable settling time, damping, and actuator movement.

Integral windup limits practical performance

If the inlet valve reaches its physical limit while a large error persists, the integral term may continue accumulating. This condition is called integral windup.

When the process becomes controllable again, the stored integral value can drive the valve excessively far, causing overshoot and delayed recovery. Practical controllers address this with output limiters, anti-windup feedback, integral separation, or temporary disabling of integration during large deviations.

Real pilot plants are not perfectly ideal

Tank friction, valve nonlinearities, transmitter limits, changing outlet conditions, and actuator saturation can make the actual process differ from the ideal (1/(As)) model.

Consequently, the zero-offset conclusion depends on adequate tuning, correct controller action, sufficient actuator capacity, and a functioning measurement loop.

How to Apply This to a Pilot Plant

The most useful distinction is whether the problem is setpoint tracking or disturbance rejection.

If your primary focus is setpoint tracking: Recognize that an ideal pure integrating level process can achieve zero setpoint offset with proportional control, although PI control may still be selected for robust behavior under realistic disturbances.
If your primary focus is load-disturbance rejection: Use integral action so the controller can establish the required new valve position without maintaining a nonzero level error.
If your primary focus is fast recovery: Tune (K_c) and (\tau_I) together, because aggressive integral action can create overshoot and oscillation.
If your primary focus is reliable operation: Include actuator limits and anti-windup protection so accumulated integral error cannot destabilize recovery.

Integral control removes steady-state offset because it supplies the persistent controller correction through accumulated error rather than requiring the process to remain away from its setpoint.

Summary Table:

Key point	Explanation
Proportional control	Requires persistent error to maintain a changed valve position after a load disturbance.
Integral control	Accumulates error and continues adjusting the manipulated variable until steady-state error reaches zero.
Liquid-level process	An ideal tank is an integrating process, so proportional control may track setpoint steps without offset.
Disturbance rejection	PI control removes residual level offset caused by changes in inlet or outlet flow.
Tuning considerations	Aggressive integral action can cause overshoot, oscillation, or sluggish recovery.
Practical protection	Actuator limits and anti-windup features help prevent excessive integral accumulation.

Need reliable hands-on control training or pilot-scale process research equipment? Contact LABPARK to discuss your requirements. LABPARK provides Educational and Vocational Unit Operations Pilot Plants for chemical engineering, bioprocess and biotechnology, and environmental and water treatment applications. Our systems help universities, research institutes, and enterprises demonstrate liquid-level control, evaluate PI tuning and disturbance rejection, train students and operators, and validate process strategies with practical, configurable equipment.