Skip to content
Pharmaceutical

Virtual commissioning pharmaceutical lines

On an aseptic line, a logic fault that surfaces after qualification is not a fix. It is three consecutive process simulations, every working shift covered, and fourteen days of incubation behind each one. We arrive with the PLC code, the sequences and the safety logic already tested against a twin of the machine.

01 The problem

Here a controls defect is not a fix. It is a restart.

The isolator is a sterility boundary: you cannot open an aseptic line and try things. Every recovery path, from a fallen vial to a restart, is PLC and robot logic. One physical test of a lyophilisation load costs a shift. And a fault found after qualification is not a correction. It is a restart.

Note

EU GMP Annex 1 §9.38: aseptic process simulation must be repeated after any significant modification to facilities, services or equipment assessed to have an impact on the sterility assurance of the product — at least three consecutive satisfactory simulation tests covering all working shifts. ISPE puts the incubation at seven days at 20–25 °C followed by a further seven at 30–35 °C.

02 Our layer

Where our work ends and yours begins.

Our record is automotive; pharma gets the method. Here "digital twin" usually means a mechanistic model predicting yield; ours is a model of the machine your PLC code runs against before the machine exists. We prove the controller, not the process. We end at the FAT and the SAT; qualification and process validation keep your quality unit's signature.

03 Where we come in

Four places where the logic decides the outcome.

  1. Aseptic filling, isolator and RABS

    Vial washer, depyrogenation tunnel, filler and stoppering run in step inside the isolator, where trying things is not allowed. Interlocks, stops under load and recovery have already run before the machine enters the room.

  2. Lyophilisation: load and unload

    A real lyophilisation cycle runs for days, so nobody rehearses load and unload on the machine. The sequence runs end to end in the model, failure cases included, as many times as it takes.

  3. Format changeover and line clearance

    Line clearance, format parts, recipe and code verification stand between batches and the first acceptable pack. It is interlock and procedural logic, the same on a vial line as on a blister line.

  4. Serialisation, rejects and exception paths

    Serialisation fails in its exceptions, not its happy flow: the reject and rework paths are what break traceability under Delegated Regulation (EU) 2016/161. Those paths have already run, not just the good one.

Note

The model we test against is a test environment in the sense of EU GMP Annex 11 §4.7, which requires documented assessments of adequacy for automated testing tools and test environments. We write that assessment, and it ships with the project.

05 What the sector itself says

Proving the control before FAT is already argued inside this sector.

  1. The sector's own engineering body published this, and attached no figures

    ISPE wrote in January 2022 that simulating plant and process during automation commissioning "results in shorter commissioning times during factory acceptance and site acceptance tests (FATs and SATs)". Our layer, named exactly; commissioning, not qualification; no percentage attached.

  2. The tools for our layer are already sold into pharma

    Siemens sells virtual commissioning for pharmaceutical machinery and names real machine builders among its customers. We are not importing this from automotive; it is already bought here.

  3. Spain's industry association describes it in the same terms

    Farmaindustria's digitalisation report describes "plant control (DCS/PLC), robotics, vision and digital twins that reduce unit costs, shorten lead times, raise OEE and cut unplanned downtime". That is the Spanish sector describing the job in its own words.

Note

A second ISPE article (May 2022) states that "digital twin technology can permit virtual commissioning" and that it "has been helpful in exposing defects early". Its authors describe the practice as emerging rather than as sector standard, and we cite it on those terms.

06 Automotive rigour

The tier your framework watches most closely is ours.

  1. We write the category that gets tested hardest

    Custom controller code is GAMP 5 Category 4 or 5: the tier your framework tests and documents in the most detail. We have written that code inside ten different OEM standards, to the programme's rules, not ours. Working inside somebody else's rulebook is precisely what is asked for here.

  2. PLC, robot and safety, one team

    The machine ships whole. Signals between PLC, robot and safety are mapped and tested together, not in two projects that first meet in the room.

  3. The people who program are the people who start it up

    The same team that proved the machine against its twin is on site through the FAT, the SAT and the start-up. No handover, no explaining the line twice. Qualification happens once.

Note

Our expertise is automotive: ten OEM standards delivered, among them VASS5 and VASS6, BMW and Daimler Integra Wave 6, with more than 400 validated PLC architectures and more than 1,000 robots programmed.

Download the case study

Request a demo