+

Three steps from
production graph
to confident decision.

Quick wins using your data to build decision certainty.

How do you increase production?

Every system has multiple ways to improve.

A bottling line — Filler, Capper, Labeler, Case Packer, Palletizer

Add Buffer?

Improve Reliability?

Automate the Process?

How do you decide?

Gut feel?

Spreadsheets?

Past experience?

Consultants?

From raw data to prediction.

Distilling 35 years of the end-to-end process into three purposely parsimonious steps.

Build

Step 1

Sketch the line and parameterize each machine's behavior.

Simulate / Validate

Step 2

Run the baseline and match history within 1% before you predict tomorrow.

Decide

Step 3

Run what-ifs. Gain ≠ Loss. Act with numbers behind you.

Step 1 · Build → Sketch

Sketch what is.

Add equipment and operational details: units, rates, conversions, and buffers or decoupling points.

"It runs on my desktop. No integration, no IT project, no waiting." — Engineering Leader, Essity

Duration: 90.00 Days Efficiency: 100.0% Perfect Production: 130,896 pallets

Source

Filler A

Capper A

Labeler A

Filler B

Capper B

Labeler B

Case Packer

Palletizer

Sink

The dice in front of you are the model.

Every interrupt has two parts. The dice represent both.

Time to Failure

How long does the machine run before something goes wrong? Roll to find out.

Time to Repair

How long until the machine is back? Roll again. That's the whole interrupt.

TTF + TTR = one interrupt. Repeat until 7 days elapse.
Five machines. When any one fails, the whole line stops.

Your handout

Roll d20 → look up minutes

d20	TTF	TTR
1	92	15
2	3	1
3	29	5
4	22	4
...	...	...
20	8	1

Same distribution, every team — just shuffled.

Mean TTF = 27.8 min · Mean TTR = 4.7 min

Same machines. Same distributions. What OEE will your team get?

Step 1 · Build → Parameterize

Compose your equipment behavior.

finds the shapes that fit each failure in your historian data.

Historical Line Event Data — blocking and starving emerge from the interaction

Full line dynamics — five machines with blocking and starving

For systems without data, the Interrupt Designer guides your distribution selection based on your reliability process.

Infant Mortality

Decreasing failure rate

Wear-out

Age-related failures

Scheduled

Fixed-duration maintenance

Step 2 · Simulate / Validate → Simulate

Simulate your baseline.

An ensemble of actors performing a thousand times.

Any duration Random seed analysis Single run Multiple runs

Step 2 · Simulate / Validate → Match history

Validate against history. Within 1%.

Model vs. historian

Interrupt Validation scatter — source vs. sim availability with 95% PI and 99% CI bands

Each point is an interrupt. On the diagonal = model matches reality.

"The only way to predict this accurately is through simulation."

— Tom Lange, 36 years Procter & Gamble

Step 3 · Decide → Experiment

What if? Gain ≠ Loss.

Eliminate each failure mode. Measure the whole system. The recovery is almost never equal to the loss.

Labeler Misalignment

Loss 6.78% → Gain 5.10%

0.75×

recover less than you lost

Filler Micro Stop

Loss 6.67% → Gain 7.97%

1.2×

56% more recovery

Same loss. Completely different recovery.

Step 3 · Decide → Act

Make confident decisions.

With the right tools in your toolbox, you can study 100 years of production in minutes.

🔧

Which failure first?

Rank by system gain, not downtime.

⚡

Faster or slower?

Faster = more product, more failures.

📦

How much buffer?

Protect the bottleneck — but how much?

🔁

Second machine?

Redundancy vs. reliability improvement.

🔄

Manual or auto resupply?

True cost is often 3× what downtime says.

🎯

The question your Loss Tree can't answer.

Every line has its own version.

AI-Powered via MCP

Ask your validated model anything.

Plain language questions. Simulation-backed answers. Powered by your specific validated production graph — not a generic AI.

🔍

Inspect topology & constraints

Node types, buffers, interrupt configs — in plain language.

▶️

Run simulations on demand

Per-constraint availability, OEE, multi-run statistics.

📈

Buffer & production analysis

Utilization, throughput distributions, constraint timelines.

22 MCP tools — works with Claude, any MCP-compatible AI, and Claude Code.

ReliaSim AI — Bottling Line Validated ✓

You

Using the Bottling Line model from Feb 1 – Mar 30, 2026 — what should I fix to improve OEE by 5%?

RS

The Capper is your critical path — 15,481 downtime events over 90 days with only 75.44% availability. Every outage starves three downstream machines. Start there.

You

Could a buffer before the Capper help?

RS

Yes — estimated +2–3% OEE at low capital cost. The buffer buys resilience; fixing the interrupts buys efficiency. Run both scenarios and compare.

You

Show me the top 5 interrupts by system gain.

RS

Top 5 by system-level gain when eliminated:

Capper Jam

7.97%

Filler Micro

5.10%

Labeler Align

3.21%

Packer Belt

2.14%

Palletizer

1.28%

Fixing the top two alone recovers +13.07% system gain.

Ask about throughput, buffers, interrupts…

+

See the impact of change before you make it.

~1%

OEE accuracy

15 min

Model construction

1,200×

Faster

300+

Organizations

"In my years at Pritsker, we came close to this, but we never made it work like this."

— Darrell Starks, Simulation Consultant, Integral Solutions

Trusted since 1995

Nestlé

AB InBev

Essity

Process Partners

Food & Bev Pharma Aerospace Energy CPG Semiconductor

Scan for the tour, podcast, and a 15-minute demo with your line.

chiaha.com

Three steps fromproduction graphto confident decision.