Correct the line while it is still on spec.
A production line rarely fails all at once. A parameter drifts, yield slips, and scrap appears shifts later. AnantState tracks each line and machine against how it should be running, so the drift is visible while a correction is cheap.
In plain termsSpot a drifting line before it makes scrap.
The shifts before the scrap
An illustration of one line, from a normal run to out of spec.
The entities are the lines, machines, batches, work orders and supplier lots a plant is run through. Each has a condition that moves with the product mix and the schedule, and occasionally with something nobody planned.
A line drifting before it scraps · illustrative
- Shift 1: Normal runOutput sits on its expected path
- Shift 4: A parameter starts to driftYield is still inside spec
- Shift 6: Cheapest point to correctAdjust now, before any scrap
- Shift 9: Out of specScrap and rework begin
- Phase: Cheap to correctfrom A parameter starts to drift to Cheapest point to correct
- Events are shown in order, not to scale.
Why it costs more than it should
- Quality is checked at the end. A line can run for shifts on a drifting parameter before an inspection finds it.
- Maintenance follows the calendar. Intervals are fixed because nobody has priced the alternative of acting on condition.
- Causes are argued afterwards. A scrap event is explained in a meeting, with no way to replay the decision that led to it.
What moves on a line
| State | Baseline drivers | Unexpected drivers |
|---|---|---|
| Line condition | Runtime, product mix, maintenance interval | Wear, contamination, operator variance |
| Yield and quality | Recipe, feedstock lot, throughput | Process drift, supplier lot variation |
| Schedule adherence | The plan, changeovers | Breakdowns, missing material |
| Unplanned-stop risk | Age, interval | Cascading failure |
A worked example
Thirty lines, and early correction on the ones drifting fastest.
Round, illustrative figures for production lines: they show the arithmetic, leave out the subscription, and are for you to change in the calculator.
| Input | Value | What it stands for |
|---|---|---|
| Production lines in scope | 30 | Everything the team is responsible for |
| Value at stake per one | $45,000 | Scrap, rework and lost output on one line over the horizon |
| Cost of the action per one | $1,800 | Correcting a drifting process parameter or inspecting the line |
| Share acted on | 30% | The lines drifting fastest: 9 of 30 |
| Modeled, USD | Pessimistic | Expected | Optimistic |
|---|---|---|---|
| Share of value at stake avoided | 8% | 20% | 35% |
| Cost to act, in total | $16,200 | $16,200 | $16,200 |
| Avoided loss | $32,400 | $81,000 | $141,750 |
| Net position | $16,200 | $64,800 | $125,550 |
Break-even is 4.0%: each action costs $1,800 against $45,000 at stake, so it has to avoid that share of the loss to pay for itself.
What exists today
Manufacturing exists as catalog templates (design to produce, equipment maintenance and others). The industrial and IoT domain compiled into the engine is the closest starting point. It is not a live simulated plant.
Bring a line that scraps
Give us a month of line events and what a correction costs. We will show where the drift was visible before the scrap.