FactorySemantics MES Open the plant
Sizing, from measured runs

What would it cost to run on your plant?

Not money: the software is free. Disk, CPU, and which database. The estimate below scales linearly from runs made on one desktop, and every constant it uses is named under the result. Then, if you want, describe the plant and I will build a scripted version of it, run it here, score it, and send you the report.

1. The plant in five numbers

Try a preset:

Machines the MES watches: state word, counters, process values.
Serialised units, packs and pallets combined. 0 if you book by count.
Raw samples are pruned on this schedule; bookings, states and ids are kept.
Every id carries an inspection record

2. Have it simulated, scored and sent back

Optional, and free. I build a scripted line from what you write, run it through a measured hour, and the MES is graded on whether it told the truth: breakdowns seen, planned stops kept out of downtime, ids recorded. You get the report and the plant description to run yourself.

Machines speak
ERP
Usually within a week; you will hear back either way.

What you write is used to build the simulated plant and to reply to you. Simulated plants are named after the process, never the company, and nothing you send is published without asking.

Estimatelinear from measured runs
Rate0
Database growth0
OPC agent0
Agent processes1
StorageSQLite
On disk0
basis · 527 B per observed id (cutlery, 57.07M ids/day → 29.7 GB/day) · 231 B per minted id without inspection · agent 23% of a core at 660 ids/s, PostgreSQL 5% · one process exact to 3× that rate, fitted ceiling 6.3× · 4.2 MB per station-day of tag samples (108-station plant, 27.5 h at 10×, 5.17 GB) · SQLite rule of thumb: under 2M ids/day and 24 stations · the runs
What "simulated and scored" means

A plant shaped like yours, with faults you named, graded.

This is the same harness the project scores itself with. It is the honest version of a sales demo: the numbers come out of your plant's shape, and the MES is marked on what it got wrong.

1 · DESCRIBE

Your paragraph becomes a line description

Stations, rates, buffers, the faults you see. fsmes sim-generate turns it into the line's tables; you get that file, and it is yours to edit and rerun.

2 · RUN

One scripted hour over real OPC UA

The simulator publishes what your machines would, the agent records it, the MES books production from counter deltas. Nothing is minted from a count.

3 · SCORE

Graded against the script

fsmes score compares what the MES reported with what the script did: breakdowns detected, planned stops kept out of downtime, seconds late per fault. You get the report and the run's database if you want it.