Introduction & Context

This reference sheet provides a framework for selecting between Feedback and Feed-forward control strategies in process engineering. In industrial systems, such as distillation columns, maintaining product purity requires mitigating the impact of measurable disturbances. Feedback control is reactive, relying on error signals, while Feed-forward control is proactive, utilizing process models to reject disturbances before they affect the output. This analysis is critical for optimizing system stability, minimizing output deviation, and managing the trade-off between implementation cost and control precision.

Methodology & Formulas

The selection process is governed by the relationship between disturbance frequency, process lag, and the Disturbance Rejection Ratio (DRR). The following algebraic framework defines the evaluation criteria:

1. Disturbance Frequency Threshold:

\[ \omega_{threshold} = \frac{1}{\tau_{col}} \]

2. Disturbance Magnitude:

\[ dist_{mag} = |z_{f,actual} - z_{f,nominal}| \]

3. Output Deviation:

\[ output_{dev,feedback} = K_p \cdot dist_{mag} \] \[ output_{dev,feedforward} = K_p \cdot dist_{mag} \cdot 0.1 \]

4. Disturbance Rejection Ratio (DRR):

\[ DRR_{feedback} = \frac{output_{dev,feedback}}{dist_{mag}} \] \[ DRR_{feedforward} = \frac{output_{dev,feedforward}}{dist_{mag}} \]

5. Thermodynamic Conversion:

\[ T_K = T_C + 273.15 \]

Parameter	Constraint / Regime
Operating Envelope	\( Z_{F,MIN} \leq z_{f,actual} \leq Z_{F,MAX} \)
Feedback Stability	If \( \omega > \omega_{threshold} \), Feedback is oscillatory
Strategy Selection	If \( \omega > \omega_{threshold} \), use Feed-forward; else use Feedback
Fluid Regime	Assumes Newtonian behavior; if viscosity varies, apply Arrhenius correction to \( \tau \)
Mixing Assumption	Assumes CSTR behavior (well-mixed)

You should prioritize feed-forward control when your process exhibits significant dead time or when measurable disturbances occur frequently before they impact the controlled variable. Use this strategy when:

The process dynamics are well-understood and can be modeled mathematically.
The primary disturbances are measurable and occur upstream of the process.
The goal is to minimize the impact of a disturbance before it causes a deviation from the setpoint.

Feedback control is inherently reactive, meaning it only takes action after an error has already been detected. Its limitations include:

Inability to anticipate future disturbances.
Potential for instability if the loop gain is set too high to compensate for slow process responses.
Delayed corrective action in systems with large transport lags or dead time.

Yes, combining these strategies is considered a best practice for high-precision process control. This hybrid approach provides the following benefits:

The feed-forward component handles the bulk of the disturbance rejection.
The feedback component acts as a trim to correct for modeling errors and unmeasured disturbances.
The system achieves faster settling times while maintaining long-term setpoint accuracy.

Worked Example: Distillation Column Feed Composition Control

A binary distillation column in a chemical separation unit must maintain distillate purity \( x_D \) constant. The feed composition \( z_F \) is subject to variations, requiring a control strategy to reject these disturbances. This example evaluates feedback versus feed-forward control based on given process parameters.

Knowns (Input Parameters and Units):

Nominal feed composition, \( z_{F,nom} = 0.500 \) mole fraction
Actual feed composition, \( z_{F,act} = 0.525 \) mole fraction
Process gain, \( K_p = \Delta x_D / \Delta z_F = 0.800 \) (unitless)
Disturbance magnitude, \( \Delta z_F = |z_{F,act} - z_{F,nom}| = 0.025 \) mole fraction
Disturbance frequency, \( \omega = 0.005 \) Hz
Column time constant, \( \tau_{col} = 300.000 \) s
Frequency threshold for feedback oscillation, \( \omega_{th} = 1/\tau_{col} = 0.003 \) Hz (from given data, rounded)
Operating temperature, \( T = 298.150 \) K (converted from 25.0°C)
Empirical design envelope for \( z_F \): \( 0.200 \leq z_F \leq 0.800 \) mole fraction

Step-by-Step Calculation:

Characterize disturbance frequency: Compare \( \omega \) to \( \omega_{th} \). Given \( \omega = 0.005 \) Hz and \( \omega_{th} = 0.003 \) Hz, since \( \omega > \omega_{th} \), feedback control would be oscillatory (as per given data: is_feedback_oscillatory = True).
Calculate output deviations for each strategy:
- Feedback strategy output deviation: \( \Delta x_{D,fb} = 0.020 \) mole fraction (from given data: output_dev_feedback = 0.020).
- Feed-forward strategy output deviation: \( \Delta x_{D,ff} = 0.002 \) mole fraction (from given data: output_dev_feedforward = 0.002).
Calculate Disturbance Rejection Ratio (DRR) for each strategy:
- DRR for feedback: \( DRR_{fb} = \Delta x_{D,fb} / \Delta z_F = 0.020 / 0.025 = 0.800 \) (unitless, from given data: drr_feedback = 0.800).
- DRR for feed-forward: \( DRR_{ff} = \Delta x_{D,ff} / \Delta z_F = 0.002 / 0.025 = 0.080 \) (unitless, from given data: drr_feedforward = 0.080).
Evaluate strategy based on oscillation and performance: Since feedback control is oscillatory (step 1) and feed-forward yields a lower DRR (0.080 vs. 0.800), indicating better disturbance rejection, feed-forward is preferred for this high-frequency disturbance.

Final Answer: The recommended control strategy is feed-forward, as the disturbance frequency exceeds the column's response capability, making feedback unstable, and feed-forward provides superior disturbance rejection with a DRR of 0.080 compared to 0.800 for feedback. This ensures constant distillate purity \( x_D \) despite feed composition variations.

"Un projet n'est jamais trop grand s'il est bien conçu." — André Citroën

"La difficulté attire l'homme de caractère, car c'est en l'étreignant qu'il se réalise." — Charles de Gaulle