A burger line with three steps, each taking 4 to 6 minutes a burger, run two ways on the same clock. For the same 300 burgers, discrete event simulates 2,993 events and discrete rate simulates 6.
- A bottling line, bulk storage to warehouse, 36 failure modes (LEDS level).
- Every machine runs at 100 bottles a minute; 100 bottles or 10 cases is about one minute of output.
- 90 days per run. Efficiency is pallets ÷ 129,600.
How these numbers were made
Every number is a ReliaSim® run, matching the ReliaSim Player run for run. One run is seed 123; the average is seeds 123 to 132.
- The same bottling line as the buffer question, 36 failure modes.
- One buffer at a time, 0 to 500 in steps of 50; the other two at zero.
- Gain: efficiency with the failure mode taken out, minus without, same buffer.
- Loss: minutes it was down in the no-buffer run (seed 123), as a share of the run.
How these numbers were made
Every bar is a pair of 90-day ReliaSim runs, matching the ReliaSim Player run for run. One run is seed 123; the average is seeds 123 to 132. Efficiency is pallets ÷ 129,600.
- A demo vegetable plant: two making lines, four vegetables, bins for each, five packing lines with their own schedule.
- Making fills the bin for its vegetable; packing draws from it. A full bin blocks making; an empty one leaves packing waiting.
- Bin lock: the bins are full of product packing doesn’t need yet, so making can’t put anything away, while packing waits for product nobody is making. Breakdowns can cause it too; in this demo it comes from the production plan.
- Two kinds of fix to try: change the plant (bigger bins) or change the schedule (making runs timed to packing).
- “Plant stuck” counts the minutes when every running making line is blocked and no running packing line is packing.
- The packing schedule is the same in every run, timed to 24-hour making runs.
How these numbers were made
Runs of the Plant Builder engine on the vegetable plant demo, one per setting. The model has no random failures, so one run is the whole answer.
Case study: The cereal plant that runs its model before every shift →
- The demo bottling line (Buffer Options, LEDS-level interrupts): five machines, every failure mode from the line’s own stop data.
- An improvement roadmap: at each step, take each remaining failure mode off the line, run it, and keep the fix that gives back the most. Repeat until the line reaches 85%.
- A fix here removes the failure mode entirely. On a real line a fix usually reduces a problem, so a real roadmap has more, smaller steps.
- Is it the optimal roadmap? On the line with no buffers we checked it three ways: a search that keeps the best 25 plans at every step instead of one, every swap of one of its fixes for another, and all 5,040 orders of its seven fixes. None beat it. Of 2,000 seven-fix plans picked at random, none reaches 85%; the best gets to 79.8%.
How these numbers were made
ReliaSim engine runs, 90 days each, the average of ten seeds, for every remaining failure mode at every step. The comparisons (every option, random plans, other orders) are engine runs too: 13,439 different sets of fixes, 134,390 runs in all.
Guide: Is 85% OEE achievable? →
- A demo steel strip line: a welder feeds strip at up to 3,000 ft a minute, a tower stores it, and a cold mill rolls it at 500 ft a minute without stopping.
- Each coil is 3,141 ft of strip. When one runs out, the welder stops to weld on the next: about 4 minutes in the model as built.
- “Starved” counts the minutes the mill stood waiting for strip.
How these numbers were made
ReliaSim engine runs of the line, 24 hours each, ten random seeds for every tower height and weld time. The page shows the average of the ten; “worst seed” is the highest of them.
Guide: How big should a buffer be? →
- SDI’s Simple SC Demo. Customer 1 orders Material 1 from Plant 1, which makes it from raw material.
- 200 days per run.
- Steady demand, then a spike.
- Customer 1’s forecast looks 10 days ahead.
- “Filled in full” counts every order so far in the run.
How these numbers were made
Saved runs of the supply chain engine the Builder runs. ×3 is the published demo; the others change only the spike size.
- Customer demand points across North America.
- Total demand.
- 16 DCs today. The engine closes one at a time and moves every customer to its best open DC.
- Cost is demand × distance, as the engine reports it.
How these numbers were made
Saved runs of the Footprint demo (greenfield engine, brownfield candidates, subset optimization). The maps are the engine’s own; open DCs in green, closed in red.