Three levers take you from measuring to controlling: monitor the process while it runs, intervene before the batch ends, and release on the measurement rather than on the wait for the laboratory.
PAT-INDUSTRY integrates in-line measurement for process control, in pharma, biopharma, chemicals, cosmetics and food. This section covers what the measurement lets you decide. Choosing the attribute belongs in Measure, choosing the technology in Technologies.
Three levers, and what each one produces
Monitor
The process signature in real time, and the endpoint of the operation.
- A more robust process
- Deviation investigations that conclude, because the trace exists
Intervene
Correct during the operation, or correct the next one.
- Yield held on the reference trajectory
- Capability that rises
Release
Replace in-process controls and release tests with the measurement itself.
- Shorter cycle time
- Lower cost of goods
Eight situations where the measurement changed the decision.
What control assumes you have already done
The three levers above rest on two prerequisites, and their order is not negotiable.
A measurement that holds. A quantity followed in line, whose real meaning is known, along with the mass of product it sees and the calibration burden it carries. That is the subject of Measure and of the technologies.
A process you understand. Knowing which parameters impact quality, and which do not. Controlling on a quantity that does not carry quality produces a loop that moves without controlling anything. That is the subject of Understand.
Without those two, a control loop is an automation, not a control strategy.
What the texts say
Process control is not a notion the PAT world invented: it is written into the reference texts, and the vocabulary matters when a dossier has to be defended.
The FDA PAT guidance of September 2004 defines a well-understood process by three cumulative conditions: critical sources of variability identified and explained, variability managed by the process, and quality attributes predicted reliably.
ICH Q8, Q9 and Q10 establish the control strategy as the object that connects process understanding to the controls actually exercised. ICH Q10 makes it part of the quality system, not a technical annex.
And the state all of this aims at has a name in the text: the state of control, which ICH Q10 defines as “a condition in which the set of controls consistently provides assurance of continued process performance and product quality”, and makes an objective of the quality system: “establish and maintain a state of control”. It is not a state you reach, it is a state you hold — which is exactly what a continuous measurement lets you observe rather than assume.
Outside pharma the requirement takes other names — capability, control plan, customer specification — and the approach does not change. The texts we work from.
Frequently asked
Do you need automation to control?
No, and that is a common confusion. The first two levers work with an operator reading a curve and deciding. Automation shortens the reaction time and removes human variability; it does not create control, it makes control repeatable. Many projects are better off starting with monitoring alone.
Can you stop at the first lever?
Yes, and it is often the right place to start. Monitoring changes neither the validated process, nor the dossier, nor the release controls: the measurement is added as an observer. It is the least binding lever, and the one that produces the knowledge on which the other two are decided.
Does the third lever require a regulatory submission?
In pharma, replacing a release test is declared and justified. Outside pharma it falls under your control plan and, where relevant, your customer’s agreement. In both cases the question belongs at scoping, not at installation.
Which lever is your process ready for?
Tell us which operation concerns you and what you measure on it today. Forty-five minutes are enough to place the lever within reach and the one that needs a prerequisite first.