Skip to content

Maintenance Service

Working Inside a Validated Facility

  • Maintenance Service
  • 2 min read

Replacing an air handling unit in an empty building is a straightforward job. Replacing the same unit while the suite next door continues to release product is a different project entirely, and the difference is almost never in the mechanical engineering.

Define the boundary before anything else

The first question is not how the work will be done but which systems and which rooms the work can affect — the requalification boundary. It determines what has to be retested afterwards, and therefore what the job actually costs and how long it takes.

Getting this wrong in either direction is expensive. If the boundary is drawn too tightly, a system that was affected goes back into service without evidence. If it is drawn too loosely, the site requalifies areas that were never at risk, which is weeks of production and testing spent proving something nobody doubted.

Impact assessment is engineering, not paperwork

A change control record is often treated as an administrative gate. It is the design document for the work. It should state which validated attributes could be affected, by what mechanism, what will be done to prevent it, and what evidence will show that it was prevented.

The mechanisms that matter are rarely exotic: a pressure cascade disturbed while a system is off, particulate released during removal or drilling, a shared control system that has to be modified, a shared duct or utility that cannot be isolated, or vibration transmitted through a structure to sensitive equipment.

Containment and sequence carry the risk

Temporary containment is a designed system, not sheeting taped to a frame. It needs its own pressure regime, usually negative to the surrounding area, its own filtered exhaust, and a defined route for people and materials that does not cross a classified boundary.

Sequence matters as much. Which system comes off first, what runs on temporary provision, when the pressure regime is broken and for how long — these decide whether adjacent areas hold their classification. This is worked out before the shutdown window is agreed, not during it.

Return to service is a plan, not an event

Reinstatement needs the same rigor: systems restarted in a defined order, the pressure cascade re-established and verified, recovery confirmed against the qualified baseline, monitoring reviewed across the whole shutdown period, and the requalification executed to the protocol agreed before the work started.

The pattern in the ones that go wrong

Almost every retrofit that overruns does so for the same reason: the requalification scope was settled after the work rather than before it. The engineering finished on schedule and the facility then spent three weeks discovering what it had to retest, against a production plan that had already assumed the area was back.

Related

More on this subject

Four checks at the basic-design gate

Before detail design opens, a GMP project should have frozen room data sheets and pressure cascade, loads that match the equipment list, a URS for each critical system and owned open items. A cascade change after the ducts are drawn reaches the duct sizes, terminal filters, AHU selection and every neighbouring pressure step.

InsightRead