Goldratt Dice Game Simulator
A hands-on simulation of dependent events, statistical fluctuations, and the Theory of Constraints.
What This Teaches
Eliyahu Goldratt introduced the Dice Game in his landmark book The Goal to demonstrate a counterintuitive truth: a production line where every station has equal average capacity will always underperform its own theoretical output. The reason is not laziness, bad management, or poor equipment, but mathematics.
Two forces conspire against every balanced line. Statistical fluctuations mean that no station produces the same output twice. Dependent events mean that a slow day at one station ripples forward through every station downstream. Gains are never passed forward. Losses always are.
The Core Insight
In a dependent system, variability accumulates. A line optimized for average performance will never achieve average throughput. The only path to reliable output is to identify and manage the constraint.
This simulator lets students experience that insight directly, then apply the Five Focusing Steps of the Theory of Constraints to improve the system and compare the results.
The Production Line
The simulator models a six-node production line: a Supplier, five Stations, and a Customer. Each day, every node rolls its dice and passes that many units downstream, but only if inventory is available. Units waiting between stations accumulate as Work in Process (WIP), visible as stacked cubes.
Each die rolls 1 through 6, giving an average of 3.5 per die per day. Management might expect 3.5 units delivered per day. The simulation shows why that expectation is wrong.
The Toolbox
Three draggable blocks sit in the top-right corner of the simulator. Drag them onto stations or the supplier to apply interventions.
⬡ Constraint
Drag to any station to identify it as the system constraint. Freely moveable as the constraint shifts.
⇢ DBR Rope
Drag to the Supplier to activate Drum-Buffer-Rope. The supplier release is capped to the constraint station output each day.
⚡ RIE
Drag to any station to simulate a focused improvement event. That station switches from rolling 1-6 to rolling only 5 or 6. One use per run.
Suggested Classroom Rounds
| Round |
Setup |
What to Observe |
| 1 — Baseline |
Use the Balanced Line preset. All stations at 1d6. Run for 20 days. |
Ask students to predict throughput before running. Compare the prediction to the result. Where did WIP accumulate? What was the Throughput Ratio? |
| 2 — Identify |
Use the Five Focusing Steps preset. All stations at 2d6 except Station 3 at 1d6. Run for 20 days. |
Students observe WIP building in front of Station 3. Drag the Constraint block to that station. |
| 3 — Exploit + Subordinate |
Drag RIE onto Station 3. Drag DBR Rope to the Supplier. Run 20 more days. |
Compare to Round 2. Throughput should improve, WIP decrease, lost revenue drop. Discuss why subordinating only the supplier is correct TOC. |
| 4 — Elevate |
Add a second die to Station 3 using the + button. Run 20 more days. |
The constraint moves. Students identify the new bottleneck. TOC is a continuous cycle, not a one-time fix. |
Reading the Metrics
Expected Throughput
Units theoretically expected, based on the average capacity of all stations that period.
Actual Throughput
Units actually delivered to the customer. Always lower than expected in an unmanaged system.
Throughput Ratio
Actual divided by Expected as a percentage. Green ≥ 80%, amber ≥ 50%, red below. The headline number for comparing rounds.
Lost Revenue
Dollar value of unmet demand at $100/unit. The cost of an unmanaged constraint made visible.
WIP Count & WIP $ Tied Up
Units sitting in queues and the capital they represent. High WIP means cash locked in the system.
Cumulative vs Period Rows
Period rows show only that window. Cumulative counts from day one, the overall health of the run.
Controls & Settings
▶Play / Pause / Step — run continuously, pause, or advance one day at a time. Step mode is ideal for walking students through the first few days.
↻Reset — clears all state, WIP, metrics, and tool placements. Does not change dice configuration.
⏸Pause every N days — the sim auto-pauses at each interval for debrief. Default is 20 days. Configurable.
±Dice +/− — add or remove dice from any station to model capacity changes. Expected throughput recalculates automatically.
⬡WIP +/− — manually adjust starting inventory in any queue. Use this to demonstrate the effect of pre-loaded buffers.