Where Lead Time Actually Goes, and How to Get It Back

Lead time is mostly waiting. A job spends a small fraction of its elapsed time being cut, formed, or assembled, and the rest sitting in a queue, waiting for the rest of its lot, or waiting behind another job on a busy machine. That is why scheduling changes lead time faster than equipment does, and it is where EDGEBIC by User Solutions puts its mechanisms: overlap the operations, shorten the waiting, and stop the sequence from wasting hours that already exist.

This post walks the four places elapsed time hides, the specific EDGEBIC mechanism that attacks each one, the arithmetic on documented worked examples, and the point at which the software stops being able to help. For the general treatment of the metric, see manufacturing lead time reduction. This post sits under the EDGEBIC results guide.

The Four Places Lead Time Hides

Take one job through one routing and split the clock four ways.

Only the first row is a shop-floor improvement project. The other three are scheduling decisions, and they are usually the larger share. That is the whole argument for attacking lead time with a schedule.

Lever One: Overlap the Operations

The most direct compression available is to stop making an operation wait for a complete lot when it only needs the first few pieces.

EDGEBIC's lot streaming supports this two ways. The piece-count model lets a downstream step start once a set number of pieces have accumulated upstream, using the documented formula: setup plus the smaller of transfer batch and order quantity, times hours per piece. The start-to-start lag model (measured in hours) suits continuous processes where a piece count is meaningless.

The worked numbers, from the book's documented examples:

Read the first row carefully, because it is the shape that matters. The turning operation still takes 52 hours. Nothing about it got faster. The job got 15 hours shorter because the driller stopped waiting for a piece it did not need.

There is a cost, and the book is explicit about it: transfer batches mean more material moves. A transfer batch of one piece is the lean ideal and it demands continuous material handling. Choose a batch size your floor can actually move, then let the arithmetic tell you what it buys.

Lever Two: Challenge the Queue and Transit Numbers

Queue time in EDGEBIC is an explicit field on the routing step, not a hidden fudge factor, and it is shift-aware. A four-hour queue starting at 14:30 on a shift that ends at 16:00 consumes 1.5 hours Monday and 2.5 hours the next working morning. It never pretends the plant runs overnight. Those handoff hours are real, and carrying queue, move, and transit time into the plan is what stops a promise date from being quietly optimistic.

That explicitness is the lever. When a number is visible, it can be defended or deleted. Most shops carry queue values that were added years ago to hide unreliable planning, and the moment planning becomes reliable, those hours are pure lead time with no purpose. See queue and transit times for how each field composes with the others.

Transit days deserve the same audit. EDGEBIC supports calendar-day and working-day modes for transit between operations, which matters when parts go to an outside heat treater: three calendar days over a weekend is not three working days, and getting the mode wrong silently lengthens every job on that routing.

One caution the diagnostics catch for you: configuring both a queue time and a start-to-start lag on the same step produces a surprise, because the lot-streaming gate replaces the queue-adjusted end rather than adding to it. EDGEBIC's anomaly checks flag that combination so the planner can confirm intent instead of discovering it on the floor.

Lever Three: Stop the Sequence From Wasting Hours

Contention is the least visible component and often the largest. In the documented three-job example (a saw and a mill, one machine each, three jobs released together), scheduling in due-date order leaves the mill idle all Monday morning while a six-hour cut blocks the saw. Total elapsed time across the three jobs: 17 working hours, finishing Wednesday morning.

Re-sequenced so the shortest cut feeds the longest mill job first, the same three jobs on the same machines finish in 14 working hours, on Tuesday afternoon. That is an 18% makespan reduction with no added capacity, and the optimizer proved it optimal on that problem (a gap of zero, meaning no better sequence exists).

Two things make this trustworthy rather than exciting. The multi-run layer is clamped never worse than your current schedule, so a run either beats the baseline or tells you it did not. And nothing is written until a planner reviews the comparison and accepts it.

Lever Four: Route to the Machine That Is Free

When several machines can do the same operation, forcing every job onto the nominal primary adds queue time that a pool would not.

Work center groups let a routing step target a pool instead of one machine, with a per-member efficiency factor so a faster machine carries fewer hours. The documented three-mill example makes the trade-off concrete: with a base of 0.04 hours per unit on 100 units, Mill-1 needs 4.5 hours, the newer Mill-2 needs 2.5, and the older Mill-3 needs 6.25. When all three are free, the earliest-completion strategy picks Mill-2 and the job finishes at 10:30. When Mill-2 is booked solid until Wednesday, the available Mill-1 wins and the job finishes Monday at 12:30 instead.

That second scenario is the point. A job that finishes Monday on a slower machine has a shorter lead time than a job that finishes Wednesday on a faster one, and only a capacity-aware pool can see that.

Putting the Levers in Order

The levers do not cost the same, so sequence them by effort:

1. Audit queue and transit first. It is a data change, it costs nothing but argument, and it often removes days.
2. Turn on lot streaming where lots are long. The formula is documented and the answer is computable before you commit to it.
3. Pool the interchangeable machines. This removes queue at the point where jobs pile up behind a nominal primary.
4. Optimize the sequence. Do this last, because it works best when the underlying data is already honest.

The heritage evidence for what this discipline reaches belongs to the User Solutions line that EDGEBIC succeeds: Technical Glass Products recorded a 4% capacity increase and a two-week lead time reduction. That is a documented outcome of the approach, not a projection for your plant.

What the Software Cannot Do Alone

It cannot shorten run time. If a part takes 1.5 hours to machine, it takes 1.5 hours. Lot streaming and sequencing move the waiting, not the cutting. Real run-time reduction is a process engineering project, and scheduling only tells you which operations are worth the project.

It cannot move material for you. A transfer batch of 20 pieces means somebody actually moves 20 pieces mid-run. Configure a batch size your material handling supports, or the schedule will be right and the floor will be late.

It cannot defend your queue times. The software makes queue values visible and shift-aware. Deciding that a two-day cooling queue is really four hours requires someone who knows the process to say so.

It cannot compress supplier lead time. Outside processes and purchased material are what they are. EDGEBIC models them as transit days and material steps so the plan is honest about them, which is useful, and that is the end of its power.

It cannot fix a routing nobody maintains. Every number above comes from hours per unit, setup, queue, and transit values on the routing. Compare a few against logged actuals before you trust a lead time computed from them.

Want to know which of the four components dominates your lead time? Bring one routing and a month of actuals to a demo, and we will split the clock four ways on your own data.