How Should Process Variables Be Classified When Designing Control Systems For Chemical Engineering Unit Operations Pilot Plants? A Practical Guide

Process variables should first be classified by their direction of influence across the process boundary. In a chemical engineering unit operations pilot plant, variables are divided into input variables and output variables. Inputs are further classified as manipulated variables or disturbance variables, while outputs are classified as measured outputs or unmeasured outputs.

The essential classification is: manipulated and disturbance variables enter or influence the process; measured and unmeasured outputs describe the process response. This distinction supports degrees-of-freedom analysis, instrument selection, and the design of feedback, feedforward, and multivariable control systems.

Classify Variables by Their Process Role

Manipulated Variables

A manipulated variable (MV) is an input that an operator or controller can adjust deliberately to influence process behavior.

Typical examples include:

Cooling-water flow rate
Steam-valve position
Heating-medium flow rate
Pump discharge flow
Reactant feed rate
Bypass flow around a heat exchanger

In a temperature-control loop, for example, the controller may manipulate steam flow to maintain reactor or process-fluid temperature at its setpoint.

Disturbance Variables

A disturbance variable (DV) affects the process but is not directly adjusted by the control system for the immediate control objective.

Examples include:

Feed flow-rate fluctuations
Feed temperature changes
Feed composition variations
Ambient-temperature changes
Utility-pressure fluctuations
Changes in upstream operating conditions

Disturbances are important because they explain why a controlled output moves away from its target. A feed-temperature disturbance in a stirred-tank heater, for instance, may cause the outlet temperature to change even when the heating-valve position is initially unchanged.

Measured Outputs

A measured output is a process result monitored directly by an instrument.

Common examples are:

Reactor temperature measured by a thermocouple
Tank level measured by a level transmitter
Flow rate measured by a flow transmitter
Pressure measured by a pressure transmitter
pH measured by an electrochemical analyzer
Conductivity measured in an electrolyte system

Measured outputs can be used as process variables (PVs) in feedback control. The controller compares the PV with the desired setpoint (SP) and adjusts an MV accordingly.

Unmeasured Outputs

An unmeasured output is a process response that is important but cannot be observed directly with the available instrumentation.

Examples include:

Internal concentration within a reactor
Composition at an inaccessible point in a distillation unit
Local temperature inside a vessel
Product quality when no online analyzer is installed
Internal stage compositions in a multistage process

These variables may be estimated from measured values and a process model. Such estimates are often called inferred variables, soft sensors, or observer outputs.

Distinguish Outputs from State Variables

State Variables Describe Internal Condition

A state variable is an internal dynamic quantity required to describe the condition of a process at a particular time.

For a multistage separation process, state variables might include:

Liquid holdup on each stage
Liquid composition on each stage
Vapor composition on each stage
Internal vessel inventories

State variables are part of the process model. They are not automatically the same as measured outputs.

Outputs Are Selected Process Responses

An output variable is a process response selected for observation, control, or performance evaluation.

For example, a distillation model may contain many internal stage holdups and compositions as state variables, while the selected outputs may be:

Distillate composition
Bottoms composition
Distillate flow rate
Column pressure

Some outputs may be measured directly, while others may be inferred. A pilot plant can therefore have many more state variables than available output measurements.

Why the Distinction Matters

The distinction determines whether a control system can regulate a variable directly or must estimate it.

If product concentration is not measured online, a controller cannot use it as a conventional feedback PV without an analyzer or estimator. Designers may instead control a related measured variable, such as temperature, pressure, conductivity, or flow ratio, and use a model to infer the concentration.

Use the Classification to Design Control Loops

Feedback Control

In a feedback loop, a measured output is compared with its setpoint. The controller changes a manipulated input to reduce the error.

A typical pilot-plant temperature loop is:

Controlled output: process-fluid temperature
Measurement: thermocouple or temperature transmitter
Manipulated input: steam or cooling-water flow
Actuator: utility control valve
Disturbances: feed temperature, feed flow, and ambient heat loss

This arrangement is effective when the output can be measured reliably and the manipulated variable has sufficient authority over it.

Feedforward Control

Feedforward control responds to a measured disturbance before that disturbance produces a significant output error.

For example, measured feed flow and feed temperature can be used to adjust utility flow in advance. Feedforward control generally requires a reasonable process model and accurate disturbance measurements, so it is often combined with feedback correction.

Multivariable Control

Pilot plants frequently contain interactions among variables. Changing one manipulated variable may affect several outputs.

Increasing steam flow may raise temperature, alter vapor generation, change pressure, and affect level. Classifying all relevant MVs, DVs, measured outputs, and unmeasured outputs helps designers identify these interactions and determine whether separate PID loops are sufficient or whether multivariable control is justified.

Match Instruments to the Classification

Instrumentation should be selected according to the outputs that must be observed and controlled.

Typical devices include:

Thermocouples or resistance temperature detectors for temperature
Differential-pressure transmitters for level or pressure
Flow transmitters for feed and utility streams
Control valves for manipulated flows
pH, conductivity, redox, or dissolved-oxygen analyzers for electrochemical variables
Spectroscopic or physical-property analyzers for composition or product-quality measurements

The measurement must be sufficiently accurate, responsive, maintainable, and compatible with the process conditions.

Apply Basic Pilot-Plant Configuration Rules

Assign One Primary Manipulated Variable per Loop

A designated process objective should have a clear manipulated variable. On a process stream between unit operations, installing multiple competing control valves can cause loops to fight each other.

The control strategy should define which loop has authority over each valve and how other objectives are coordinated.

Control Phase Interfaces with Level Loops

Whenever a gas-liquid or liquid-liquid interface must be maintained, a level controller is normally required.

The controller manipulates an inlet or outlet flow so that inventory remains stable. This is particularly important in vessels, separators, reboilers, condensers, and other equipment where phase holdup affects safe and predictable operation.

Place Liquid-Level Valves Appropriately

For a typical liquid-level control arrangement, the control valve is generally placed on the pump discharge line rather than the suction line.

This preserves suitable pump suction conditions and allows the valve to regulate the discharged liquid flow without unnecessarily compromising pump hydraulics.

Use Utility Flow or Bypass Flow for Temperature Control

Temperature is commonly controlled by adjusting the flow of a heating or cooling utility.

An alternative is to use a bypass around a heat exchanger, allowing the controller to vary the fraction of process fluid that receives heat transfer.

Establish the Main Material Balance at the Feed

The main feed flow regulator or ratio controller often establishes the material input to the pilot plant.

This creates a stable basis for downstream level, flow, composition, and residence-time control. Feed disturbances should still be identified separately if the feed is not fully controlled.

Understanding the Trade-offs

Measured Variables Improve Feedback but Add Cost

Direct measurement generally gives the controller better information about the actual process response.

However, analyzers can be expensive, slow, difficult to calibrate, or demanding to maintain. For educational pilot plants, the instrument must justify its cost through its training or research value.

Unmeasured Variables Require Estimation

Estimating concentration or internal composition can reduce the need for additional analyzers.

The limitation is that an inferred output depends on model quality, sensor reliability, and process assumptions. Estimation should not be treated as equivalent to direct measurement when the consequences of error are significant.

Too Many Loops Can Create Interaction Problems

Adding a controller for every available measurement does not automatically improve control.

Closely coupled loops may compete for the same manipulated variable or amplify oscillations. The control structure should reflect process causality, available actuator authority, and the dynamic response of each unit operation.

Classification Does Not Replace Degrees-of-Freedom Analysis

Labeling a variable as manipulated or a disturbance does not by itself establish whether the plant is controllable.

The designer must still compare the number of independent manipulated variables with the required specifications and constraints. This confirms whether the proposed control objectives can be achieved simultaneously.

Educational Visibility Must Be Balanced with Industrial Realism

A pilot plant should expose students to meaningful PV, SP, and MV relationships, trends, and operating responses.

At the same time, excessive instrumentation can obscure the process and increase maintenance requirements. The best design makes key disturbances and control actions visible without instrumenting every possible variable.

How to Apply This to Your Pilot Plant

Begin by listing every variable that crosses the process boundary or influences the unit operation, then assign each variable a control role.

If your primary focus is degrees-of-freedom analysis: Separate adjustable manipulated inputs from independent disturbances, and count only genuinely independent variables.
If your primary focus is feedback control: Identify the measured outputs that can serve as reliable PVs and pair each with an appropriate manipulated variable.
If your primary focus is product quality: Treat composition and concentration as controlled outputs, then determine whether they require direct analyzers or state estimation.
If your primary focus is disturbance rejection: Measure important disturbances such as feed flow, temperature, or composition so that feedforward action can be evaluated.
If your primary focus is multivariable behavior: Map how each manipulated variable affects every important measured and unmeasured output before assigning independent PID loops.
If your primary focus is safe pilot-plant operation: Give priority to level, pressure, temperature, and flow measurements that protect equipment and maintain stable inventories.

A well-designed classification turns a collection of sensors and valves into a control structure whose behavior can be analyzed, taught, and improved.

Summary Table:

Variable classification	Role in the process	Typical examples	Control-system use
Manipulated variable (MV)	Adjustable input used to influence process behavior	Steam flow, cooling-water flow, pump discharge, feed rate	Controller output and actuator selection
Disturbance variable (DV)	Uncontrolled input that affects the process	Feed temperature, composition, flow changes, utility-pressure fluctuations	Feedback analysis and feedforward control
Measured output	Process response monitored by an instrument	Temperature, level, pressure, flow, pH, conductivity	Process variable for feedback control
Unmeasured output	Important process response not directly monitored	Internal concentration, stage composition, product quality	Estimation, soft sensors, or model-based control
State variable	Internal quantity describing the process condition	Holdup, vessel inventory, internal composition	Dynamic modeling and observer design

Planning a pilot plant that demonstrates real control-system behavior? Contact LABPARK to discuss your requirements. LABPARK provides Educational and Vocational Unit Operations Pilot Plants for universities, research institutes, and enterprises, covering chemical engineering, bioprocess and biotech, and environmental and water treatment applications. Our systems help you connect sensors, manipulated variables, disturbances, measured outputs, and control loops in a practical training or research environment. We can support your team in selecting suitable instrumentation, configuring feedback and feedforward strategies, evaluating multivariable interactions, and balancing educational visibility with industrial realism. Contact LABPARK to develop a reliable, application-focused pilot plant for teaching, process research, workforce training, or enterprise demonstration.