Virtual commissioning is the practice of running the actual PLC program, ladder logic, structured text, and robot code against a 3D physics-based digital twin before any hardware reaches the floor, then exchanging real I/O signals at PLC scan rate through a co-simulation interface [S1][S2].
For a process engineer weighing whether to fund a virtual commissioning cell, the operational answer is yes for most safety logic, interlocks, and state-machine faults, with three caveats: the model must be functionally accurate, the safety controller's certified firmware must be emulated (not approximated), and any SIL-rated hardware fault injection still needs bench testing.
What "Safety Logic" Virtual Commissioning Actually Validates
Virtual commissioning validates the same software a programmable logic controller will execute in production, against a kinematically accurate 3D replica of the machine, robot, conveyor, and sensor suite, exchanging closed-loop I/O at scan rate [S2][S4]. The validation includes sequential logic, interlocks, state machines, error handling, and HMI integration, and the issue classes typically surfaced are race conditions, deadlocks, incorrect state transitions, and missing error handlers [S2]. Per the March 2026 ifactoryapp analysis, 70% of commissioning delays trace to errors in control software, not mechanical rework or supply issues [S2].
The scope of what is testable is wider than most specifiers assume. Engineers can force error conditions, trigger rarely used states, and time-warp weeks of wear or drift into hours of simulation [S3]. Safety routines and emergency stop sequences can be run without exposing personnel or equipment to the real fault, which makes the virtual cell function as a dress rehearsal for the machine safety envelope before site power-on [S3][S5].
What the Co-Simulation Loop Looks Like
Three components define a virtual commissioning cell: a digital twin model imported from CAD with kinematic, dynamic, and logical parameters; the actual controller code executed via virtual PLC emulators (Siemens PLCSIM Advanced, Rockwell Studio 5000 Logix Emulate, Mitsubishi GX Simulator) plus manufacturer-native robot simulators (FANUC Roboguide, ABB RobotStudio, KUKA.Sim); and a real-time co-simulation interface that exchanges I/O at scan rate so the virtual model sends sensor signals to the PLC and receives actuator commands back [S2].
The closed loop is what separates virtual commissioning from 3D animation. Per Tecnomatix guidance, virtual commissioning allows one to debug PLC code in a virtual environment before downloading it to real equipment, exercising interlocks, sequences, and safety checks against the real control logic [S1]. The iFactoryApp March 2026 dataset pegs ROI on the first project between 3:1 and 5:1 [S2].
Decision Criteria: When Virtual Commissioning Pays vs When It Does Not
Virtual commissioning pays when the project has PLC plus HMI plus one or more kinematic assets (fixtures, conveyors, rails, robots, NC machines), which is the profile Tecnomatix identifies as the threshold for a justified cell [S1]. It does not pay for simple relay panels, single-axis machines, or for projects where the safety functions are implemented entirely in a hardwired safety relay with no software path to validate.
A criteria-based comparison lines the main options up against four decision factors:
- Traditional on-site commissioning only: lowest tooling cost, longest critical-path exposure, 70% delay risk concentrated in the first 2 AM of PLC debug, no advance signal [S2].
- Static 3D review (no PLC hookup): catches reach and collision geometry only, misses all logic and timing faults, modest cost [S4].
- Full virtual commissioning with virtual PLC: catches logic, sequence, collision, and interlock faults, requires emulation licensing and a 3D model with full kinematics, the standard buy for complex cells [S1][S2][S5].
- Virtual commissioning plus real safety controller on a bench: adds SIL-rated hardware fault injection and is required where the safety function depends on certified firmware behaviour, watchdog timing, or dual-channel cross-check that the virtual PLC cannot fully reproduce [S2][S5].
Failure Modes Virtual Commissioning Can and Cannot Catch
Caught virtually: software race conditions, deadlocks, wrong state transitions, missing error handlers, robot reach singularities, fixture collisions, suboptimal motion paths, and HMI logic that does not match the PLC's state map [S2]. The model also catches sensor wiring inversions and interlock polarity errors because the co-simulation returns the same Boolean the real I/O card would return [S2][S8].
Not caught virtually: real-world sensor noise, contact bounce, encoder jitter, hydraulic compressibility, thermal drift, EMC-induced faults, and the actual response time of a hardwired safety barrier or e-stop relay. These still require bench test or controlled site energisation, per TTS Network's staging recommendation that on-site commissioning remains a phase even after the virtual cell passes [S4][S5]. Per the Visual Components treatment, a textbook definition calls virtual commissioning "the act of setting up, testing, and validating a system's design and operation in a virtual environment on a computer," and notes that most providers retain a real commissioning phase to prove out machine operation [S4].
Standards, Compliance, and the FAT/SAT Question
Virtual commissioning aligns with factory acceptance testing (FAT) discipline and, per Visual Components, in some cases enables a Virtual FAT where the test plan is executed against the digital twin before the hardware is shipped [S4]. For projects bound by fire safety or functional-safety codes, the virtual cell supports the verification documentation but does not replace the SAT (site acceptance test), which is signed against the installed hardware [S4].
For integrators, this shifts the bulk of the debug loop off the critical path, in parallel with mechanical build [S4].
Trackable Signals to Watch
iFactoryApp reports that virtual commissioning can eliminate 15-25% of on-site troubleshooting costs [S2], and Visual Components positions the practice as a way to not just hit the FAT date but pass the FAT [S4]. For adjacent reading, a recent spec-side look at generative AI for PLC code covers where AI-generated logic adds risk to that same virtual cell, and AI maintenance scheduling shows the operational twin that the validated model can feed once the line is live.