What Is the Delay Composition Order in Scheduling?

The delay composition order is the fixed sequence in which a scheduler applies a routing step's between-step delays after its operation finishes: queue time first, then any lot-streaming overlap, then transit days, producing one timestamp that is the earliest moment the next step may begin. The order is not cosmetic. Two of the three delays add to what came before, while the middle one replaces it, so knowing the sequence is the difference between predicting a handoff correctly and guessing. EDGEBIC by User Solutions runs the same three stages in the same order on every non-material step, every run.

How it works

When an operation is placed on the calendar, the engine knows when the machine work ends. That end time is rarely the moment the next step can start, because a routing can carry several kinds of deliberate delay. Three stages resolve them.

Stage one, queue time. The step's queue time is added to the operation's end, counted in shift hours rather than clock hours. A two hour queue that begins at 15:00 on an eight hour day shift consumes the one hour left in that shift and then picks up the next working morning. If queue time is zero the stage returns immediately and the clock does not advance at all, not even past a holiday.

Stage two, lot-streaming overlap. If the step carries either a flow lag in hours or a transfer batch in pieces, the streaming calculation runs. This is the stage that behaves differently from the others: its result replaces the value coming out of stage one rather than adding to it. Streaming can therefore pull the downstream start earlier than the operation's own end, which is the whole point of overlapping two stations, or push it later, depending on the setting. If neither streaming field is set, the stage is skipped and stage one's value passes straight through.

Stage three, transit days. Unless the step is a material step, transit days are applied from wherever stage two left the clock. Transit advances in whole days and you choose whether those days count as calendar days or as days the work center actually operates. Material steps skip this stage, because a material step already absorbs its transit internally when its own end date is built, so applying it again would double-count.

The composed timestamp is then published as the step's end time, and the next step reads it as a lower bound. It is an eligibility marker, not a placement. If it lands on a Saturday night, the allocator simply searches forward from there for the first shift slot with room on the correct work center.

A concrete example

A plating operation feeds an inspection step. The plated parts need one hour on a wet rack to drip, then a one calendar day trip to an inspection lab in another building. The shift is Monday to Friday, 08:00 to 16:00. The plating operation finishes Friday at 15:00.

Inspection's earliest start is Saturday 16:00. No Saturday shift runs, so the allocator searches forward and places inspection Monday at 08:00.

Change one setting and watch the arithmetic move. If transit is counted in working days instead of calendar days, Friday is day zero and the next working day is Monday, so stage three ends at Monday 16:00. Inspection cannot begin before then and lands Tuesday at 08:00, a full day later than the calendar-day version. Neither answer is wrong; they model different real-world delays, which is why the choice exists.

Now add a flow lag to the plating step and leave the queue time in place. Stage one still computes Friday 16:00, but stage two overwrites it with the streaming result, and stage three works from that instead. The drip time you configured has quietly stopped contributing. That is the interaction most worth remembering.

How EDGEBIC uses it

Each stage has its own field on the routing step, and each has its own definition worth reading separately: queue time in manufacturing scheduling for the shift-aware buffer, what is lot streaming and what is flowstep overlap for the middle stage, and what is transit time in production scheduling for the trip.

The calendar-versus-working-day choice inside stage three has its own consequences, laid out in working day vs calendar day in scheduling. And the reason material steps skip stage three entirely comes down to how those rows are built, which is covered in what is a material step in a routing.

Practically, the order gives you a diagnostic recipe. When a gap between two operations looks wrong, work the three stages in sequence rather than staring at the whole gap. Confirm the queue time, check whether a streaming field is set and therefore overriding it, then confirm the transit count and its day mode. One of those three almost always explains the gap, and the order tells you which to suspect first.

The takeaway

Three delays, one fixed order, and one of them overwrites instead of adding. That single asymmetry is why a routing step with both a queue time and a flow setting does not behave the way most planners expect, and why the composition order is worth knowing by name rather than by trial and error. Read it as the handoff arithmetic between two operations: eligibility first, placement second. For the individual pieces see what is queue time in manufacturing scheduling and what is transfer delay in lot streaming, then explore EDGEBIC or, if you are migrating from the legacy product, RMDB to EDGEBIC.