Critical Ratio in the EDGEBIC Optimizer: A Pace-Aware Sequence

EDGEBIC by User Solutions does not sort by critical ratio by default. It uses critical ratio as one candidate ordering inside the multi-run optimizer, which builds a complete finite schedule for each candidate, scores them, and proposes the best, never worse than your baseline. You can also apply critical ratio by hand through the priority integer or start date. This article explains both paths and when to use each.

Quick refresher

Critical ratio is time remaining until the due date divided by work remaining. Below 1.0 a job is behind pace, at 1.0 it is on pace, above 1.0 it has slack, and the lowest number runs first. The rest of this article assumes that much; the full explanation is in the critical ratio guide.

Path 1: the default engine, and why critical ratio is not in it

The standard engine sorts every order deterministically by priority, then start date, then due date, so the same inputs always produce the same sequence and any change traces back to a data change. That determinism is a feature, and it is described in how a scheduler decides which job runs first.

Critical ratio is the opposite kind of rule: it changes on its own as days pass and as operators log completed hours. A job that looked safe on Monday can slide under 1.0 by Wednesday if an upstream step ran long. That is exactly what makes it useful, and exactly why it does not belong in the everyday default sort, where a plan that reshuffles itself without a data change would be hard to trust.

Path 2: the optimizer, where critical ratio competes

Critical ratio enters through the optimizer. The multi-run search re-runs the unmodified engine under several complete global orderings, including an earliest-due-date ordering, shortest and longest processing time, priority-then-due-date, seeded shuffles, and a critical-ratio ordering. It scores every resulting schedule and proposes the single best one. That proposal is clamped never worse than the baseline, and nothing changes until you accept it.

The point is that sequencing rules interact with capacity. In a finite capacity plan a machine takes one job at a time, so the order decides who is late, but a rule that looks right in isolation can lose once real shift hours, setups and shared machines apply. Instead of betting the plant on one dispatch rule, you let the optimizer test critical ratio as one candidate among many and keep it only if it genuinely helps. The mechanics live in the optimizer guide, and the goals that decide what "best" means are in the optimizer goals and presets guide.

A small example

Three jobs compete for one mill on Monday; Friday is 4 working days out. MO-201 (due Friday, 2 days of work) has a ratio of 2.00, MO-202 (due Friday, 6 days of work) 0.67, and MO-203 (due next Tuesday, 5 days of work, 6 working days out) 1.20. The critical-ratio candidate orders them MO-202, MO-203, MO-201. The baseline sort, with equal priorities and start dates, falls back to due date and puts the two Friday jobs ahead of MO-203 in their existing order. If MO-201 happens to be first in that order, the baseline spends the first two days on the one job with margin. The optimizer builds both complete schedules, scores them, and the critical-ratio candidate wins if it cuts lateness under your goal.

Path 3: applying critical ratio by hand

Because critical ratio is just time remaining over work remaining, you can compute it from the same fields you already maintain (the due date and the routing hours) and use it to inform how you set the priority integer or the start date. The engine then honors your ordering directly. This is useful for a single hot job, or for a planner who wants to review a pace-aware order before the optimizer runs.

When to leave it out

On a heavily setup-dependent line, running strictly by critical ratio can scatter similar jobs and pile up changeover time, which is why the optimizer also weighs a setup-minimizing sequence; that trade-off is covered in how a scheduler trades off setup against due date. And if routing hours are wrong, the ratio is wrong. Clean routing data is the price of admission, in a job shop especially.

See how a pace-aware sequence would reshape your own queue in EDGEBIC, and read the full sequencing pipeline in the scheduling engine guide.