Digital Twin Series · Level 5 · Case study
Digital Twin Case Study: An Automated Machine
What does one digital twin do for one machine, from the first CAD model to live monitoring, and what can I test myself?
This case follows an automated battery cell stacking machine: a robot takes cells from a conveyor and builds stacks of four, with a target of 12 seconds per stack. The same twin is used to design the machine, check the cycle time, test the PLC program, commission it virtually, watch it live and try changes before they reach the line. Each step below has a working model of this machine. The figures are an illustrative case, not measured data: the tools here compute their own numbers, and you can put in yours.
A stack is built from 4 cells. A cell arrives every pitch (0.3 m) of conveyor travel, the robot needs about 1.1 s to pick and place one cell at full speed, and the stack then takes a fixed time to clamp and transfer. Each cell slot lasts as long as the slower of the conveyor feed and the robot. Speeding up the conveyor helps only until the robot becomes the limit. The pitch, robot time and transfer time are assumptions of this teaching model; take yours from the CAD model and the robot path simulation.
This is a small virtual PLC for the stacking sequence: wait for a cell, pick, place, repeat until the stack is full, then clamp and transfer. Switch the faults on while it runs: an open guard door stops the machine at once and needs a reset, a missing cell raises an alarm after a timeout, and a vacuum loss gets one retry before it alarms. Testing fault and safety logic like this before the real machine exists is what virtual commissioning is for.
Mechanical installation and wiring cannot be done virtually. Sequence, interlocks, fault handling, timing and the HMI can, but only as far as the model matches the real machine: whatever the model gets wrong has to be re-checked on site. The task times and the shares are assumptions; use the record of your own last commissioning.
A simulated machine feeds the board. The twin compares the live cycle time with the target and with what the model expects, and raises an alert when the machine drifts away. The live state also lets you start what-if runs from where the machine really is.
| Change | Cycle time (s) | OEE (%) | Result |
|---|
OEE here is availability x performance x quality, with an assumed breakdown loss, micro-stops that fall as the buffer grows, a fixed best cycle time and a fixed first-pass yield. The base case is 10 m/min, 3 operators, a buffer of 2 and cells 100 mm wide. These relations are assumptions of the teaching model.
The starting values are the illustrative case from the infographic, not measurements. Replace them with the figures you measured before and after on your own machine. Savings are counted only from stops, scrap and extra output that has demand behind it. A one-time saving from shorter commissioning is added once.
| Stage | Tools shown in the case | What they do here |
|---|
The tools are examples, not recommendations. Choose by what your team owns and what your machines and PLCs support.
The case is illustrative and all simulators are simplified teaching models. Measure your own machine, validate your own model and review safety logic with a qualified safety engineer.