Simulate the kaizen before you move the racks
The kaizen team agrees to merge two buffers and tighten the milk run. Facilities starts marking the floor before anyone has tested whether the new pitch creates a new pinch at the paint aisle door.
Why lean still needs a what-if step
Lean manufacturing rewards small experiments, yet some countermeasures move racks, add automation, or freeze a new heijunka sequence that touches every supplier call. Doing those live first is how plants lock waste into concrete. A simulation twin, often built with discrete-event simulation, lets teams trial mix, buffer size, and route options before they buy the equipment or commit the layout.
Digital lean case work shows the same pattern: simulate bottlenecks and schedules, then measure. Plants that combine lean methods with twin-based simulation report large productivity swings when hidden queues and outsourcing dependencies finally become visible in the model. Treat those outcomes as motivation to simulate, not as a promise that your hall will match another site's percentage.
Simulation twin versus operational twin
The simulation twin answers what-if. The operational digital twin answers what-is. Use simulation to compare leveled sequences, automated guided vehicle (AGV) / autonomous mobile robot (AMR) counts, or rack moves. Use the live hall after go-live to see whether empty travel, dwell, and starves actually fell. Skipping either side recreates a familiar failure: a perfect future-state slide, then a week of surprises.
Visibility before automation remains the lean logistics rule. Automating a broken milk run copies muda into CapEx. Measure residual load on manned movers first, then decide which corridors are stable enough for an automated guided vehicle (AGV) / autonomous mobile robot (AMR). That decision path sits under lean logistics under Industry 4.0 is visibility before automation and why RTLS projects usually start with logistics.
After live runs exist, keep them in a spatially oriented dataset. Correlation and causality analytics can test whether the simulated pinch still shows up as a real pairing on busy shifts. Continuous improvement needs both the paper trial and the stored reality.
Heijunka and layout on the same decision table
Heijunka levels volume and mix so people and assets avoid overload waves. Leveling on a board while the hall still surges is only half the job. Simulation can test whether a finer mix overloads a shared prep cell. The operational twin can show whether material call-offs still arrive in clumps after the schedule looks smooth. Faster changeovers from SMED often unlock finer leveling. Simulate that unlock before you promise sales a mixed sequence the assets cannot hold.
Simulation cost framing sits under what a simulation digital twin costs and what it buys and simulation digital twins test the change on paper.
What a good simulation input looks like
Heijunka trials deserve special care. A leveled mix that looks smooth in average demand can still spike a shared prep cell at the top of each hour. Simulate peaks, not only means. Then watch the live hall for the same hourly wave after go-live.
Simulation is only as honest as the constraints you feed it. Optimistic changeover times, ignored aisle width, and average demand that hides peaks will bless a layout the live hall cannot run. Pull real cycle times, real setup distributions, and real traffic peaks from the operational record when it exists. When it does not exist yet, instrument a short baseline before you trust a future-state model enough to move racks.
Good lean simulations also name the decision they support. Buffer size for Station 4. Mix pitch for paint. Number of tuggers for a milk run. automated guided vehicle (AGV) / autonomous mobile robot (AMR) count for a stable corridor. Vague optimize the factory models burn months. Narrow decisions finish in time to affect this quarter's kaizen budget.
After the physical change, resist the urge to declare victory from the simulation report alone. Re-measure empty travel, dwell, and starve counts on the live twin for enough shifts to cover mix variation. If the live hall disagrees, update the model and the standard. That closed comparison is how simulation earns a permanent seat in lean, instead of a one-off consultant phase.
How teams sequence simulate then measure
1. Write the lean countermeasure as a testable change - Buffer move, mix sequence, or route rule.
2. Simulate against honest constraints - Changeover times, aisle width, crew size, and peak demand.
3. Pilot on the floor only after the model beats the current state - Bound the pilot area.
4. Re-measure on the operational twin - Standardize only what live paths and dwell confirm.
Archive the simulation assumptions next to the live after-report. Future teams need to know what demand, changeover, and aisle constraints were believed. Without that archive, every new layout kaizen rebuilds faith from zero.
twinzo focuses on the live proof side through internal logistics optimization and production monitoring. Get in touch if you want to compare a planned layout kaizen with live hall evidence before the racks move.