Trexin ADM Winter Haven / Digital Twin · Concept Demo
Simulated data · Illustrative only
The art of the possible

Move the levers. Watch the plant re-plan itself.

A living model of the WIP floor: pick what you need to make, set your priorities, and the twin proposes the run — which drums to pull, which still, how many passes — with the reasoning shown. What takes 4–8 hours today happens as fast as you can move a slider.

Levers

Mass required
From the distillation request sheet.
Optimize for
Fresh oil allowed
Cap on fresh cold-pressed oil in the feed. Lower = more WIP consumed.
Spec confidence floor
How sure the plan must be before it stops adding passes or better feed.

The WIP floor

Terpene-rich Aldehyde cuts Heavy ends Mixed returns Uncoded (legacy) Pulled for this run

Each square is a lot on the floor — 0 drums light up as the twin selects feed. In production this grid is the real 4,500-lot inventory from the daily batchmetrics dump, and every selection carries its GC profile, weight, and age.

Proposed run plan

Predicted feed GC vs. target windowsignal / retention
Blended feed (predicted) Target spec window
Planning time
<1 min
vs. 4–8 hrs by hand today
WIP consumed
First-pass confidence
baseline ~85% today
Fresh oil needed
constraint: minimize
Still & passes

The compounding effect

If plans like this one run out daily, here is the pace toward the 2026 WIP-reduction target. This is the lever leadership cares about — every slider above changes this line.

$0$8.5MM$17MM target

“Make me something and I’ll sell it”

The reverse question — the one nobody has time to answer by hand. The twin scans every drum on the floor and ranks what the current WIP could become, so sales can go sell it before procurement buys another pound of fresh oil.

In production, each card links to a one-click run plan (the “Plan a run” view, pre-filled) and shows real sales pricing pulled from JDE. Values here are simulated to show the shape of the answer, not the answer itself.