ReliaSim / OEE simulation

OEE Simulation: Predict OEE Before You Change the Line
From per-machine interrupt data, validated before it predicts

OEE measured on the line is a record of what happened. It cannot tell you what a buffer, a faster line speed or an eliminated failure mode will do to it. OEE simulation can: model the line from its own interrupt data, check the model against the historian, then change one thing and read the predicted OEE.

Why measured OEE cannot predict a change

OEE multiplies three factors — Availability × Performance × Quality — and grades execution against planned production time. It is a good scorecard. As a forecasting tool it has three structural problems, none of which more care with the data can fix.

• It has no history for the change. The questions worth spending money on are counterfactual: a buffer you have not installed, a speed you have not run. No trend line on last year’s OEE contains that line.
• Losses interact. Ten minutes down on a machine feeding a full buffer costs nothing; the same ten minutes on a starved constraint costs ten minutes of finished product. The OEE report records both identically.
• Averages run optimistic. For machines in series with buffers between them, expected throughput is strictly less than the figure you get by multiplying average rates and availabilities. The error runs one way, so spreadsheet forecasts of OEE come out predictably too high.

So when an improvement project promises a number of OEE points, the number usually came from subtracting a loss from a loss report. What actually comes back is often different, in both directions — the subject of which loss to fix first.

Machine OEE is not line OEE

The world-class benchmark of 85% is a single-machine figure. Carry it to a line and topology decides what it is worth:

The real row is the one that matters, and it is the one no formula reaches. Multiplying machine availabilities assumes the machines fail independently; buffers are precisely what break that assumption. That is why a plant that multiplies out to 46.7% can measure 54.3% — the arithmetic cannot see the decoupling the physical line actually has. Is 85% OEE achievable on your line? works through the consequences.

How OEE simulation works

In ReliaSim, each factor of OEE lands somewhere concrete in the model:

The interrupt data is the part that decides accuracy. With a Line Event Data system or a historian, ReliaStats fits each failure mode’s distributions from the raw events — downtime data analysis is its job — and hands ReliaSim parameters it reads directly. For a line that does not exist yet, the Interrupt Designer lets you enter the shapes by hand.

Named failure modes matter more than they look. One lumped downtime figure per machine still gives a working model, but a flat answer: in Fischel and Lange’s words, the gains and losses come out “less accurate and essentially identical to each other.” Per-mode detail is what lets the model tell one improvement from another.

Then the engine runs the line forward through time — a year, or many years across replications — and reports the efficiency it produced, per machine and for the line. Because the result is a distribution rather than a single figure, you see the spread a single year of history can hide.

Validate against measured OEE first

A simulated OEE is worth nothing until the model has reproduced the OEE the line actually delivered. The standard ReliaSim holds to is per failure mode, not merely in aggregate: every interrupt’s simulated availability is compared with its observed availability, and each should land on the match line inside its confidence band. A single overall number can be right for the wrong reasons — one mode overstated, another understated, the errors cancelling. A per-mode comparison cannot hide that.

The evidence that the method works at plant scale is published. In a 2020 Winter Simulation Conference paper, Fischel and Lange modeled a multi-line food plant — over twenty unit operations, up to twenty failure modes each — and set their validation bar at agreement within 1% of overall system OEE between a one-year simulation and the plant’s actual data. It met it. That model was built in ExtendSim; it was rebuilt in ReliaSim and independently validated by Tom Lange within 1% of both the plant’s measured OEE and the ExtendSim model, running roughly 1200× faster on the same laptop. The full account is in the published OEE validation case study.

Simulated efficiency and your reported OEE

Put a simulated figure next to a plant dashboard and they may differ by twenty or thirty points. Usually neither is wrong. OEE divides by planned production time; ReliaSim’s efficiency divides by every minute of the modeled period, staffed or not — the stricter, every-minute convention. On a plant that runs 24/7 the two denominators coincide and the numbers should agree; on a one-shift plant they part company by exactly the time left unscheduled.

The reconciliation is simple: multiply the scheduled-time figure by the share of calendar time you actually scheduled, and any remaining difference is genuinely about the model. Asset efficiency vs OEE covers the arithmetic, and how big is your hidden factory? covers what the gap is worth.

What you can predict

Once the model reproduces the line, every question becomes the same experiment: change one thing, re-run, compare.

• Remove a failure mode. The OEE that comes back is not the loss the report recorded. On the demo bottling line, the Labeler misalignment mode costs 6.67% efficiency but returns 4.85% when removed — a buffer was already absorbing part of it. Others return more than they cost, because they were cascading downstream.
• Resize a buffer. Sweep capacity and see where the OEE curve flattens — how big should a buffer be?
• Change line speed. Running faster raises failure rate too, so OEE and output have an optimum rather than rising with speed.
• Find the ceiling. Turn every interrupt off and the model reports the most this topology can produce, so you know how much OEE is actually available to recover.

Deciding which of those improvements to fund, and what each trades away, is its own question — see OEE improvement tradeoffs. This page is about the instrument that prices them.

Runs are cheap enough to do this exhaustively. The buffer sweep behind the buffer guide is 7,770 runs — 1,864 simulated years in 31.8 seconds — so you compare every candidate rather than the three someone nominated. It all runs in ReliaSim’s manufacturing simulation software on your own desktop, and the discrete rate simulation engine is what makes that speed possible.

Frequently asked questions