Spotting a Bottleneck on the EDGEBIC Gantt

A bottleneck on the EDGEBIC Gantt is the lane that is full when the others are not. Every operation is a bar on a work center lane, so a constraint shows itself the way it does on the floor: one machine's row is packed wall to wall, day after day, while the lanes around it have gaps. Work stacks up to reach it, and when you touch it, the change ripples across everything behind it. EDGEBIC by User Solutions puts every job on one timeline so this pattern is visible at a glance instead of hidden in a report.

This post is a field guide to reading a constraint off the Gantt, confirming it, and telling the engine to respect it. For the broader picture of visual scheduling, see the visual scheduling guide, and for the concept itself, see production bottleneck identification.

The Visual Signature: A Packed Lane

The Gantt arranges work into lanes, one per work center, with each operation drawn as a bar in its machine's lane. Read the lanes vertically and a bottleneck announces itself. The constraint lane is a dense, continuous run of bars with no daylight between them. Its neighbors are looser, with visible gaps where a machine sits idle waiting for parts.

The reason is simple. A bottleneck is the slowest link in the chain, so work arrives faster than it can leave. On the Gantt that arriving work has to go somewhere, and it queues up as back-to-back bars in the constraint's lane. The machines feeding it finish early and wait; the machines behind it start late because they are starved. The full lane in the middle is the constraint.

One packed day is not proof. A machine can be busy on Tuesday and idle on Wednesday. The signature of a real bottleneck is a lane that stays packed across the whole visible horizon, run after run. Widen the timeline with the horizon commands and look again. If the lane is still solid a week out, you have found the constraint, not a busy afternoon.

The Cascade: How Far a Machine Reaches

There is a second, more dynamic signal, and it appears the moment you change something. When you move an operation, EDGEBIC marks the later operations in the same job that now start after your moved bar's new end. They turn to the Downstream Changed color, an amber "these may be out of sequence" flag. It writes nothing to the database; it is a review signal.

Move an operation on an ordinary machine and a couple of downstream bars light up. Move an operation on the constraint and a dozen light up, across several jobs, because so much of the plant depends on that one machine. The size of the cascade measures the machine's reach. A constraint is, almost by definition, the machine whose changes ripple furthest. Watching the amber spread is a live confirmation of what the packed lane already suggested. This is the same ripple effect covered in how drag and drop shows the downstream ripple.

Confirming It: From Eyeball to Number

The packed lane and the wide cascade are fast, honest signals, but you can confirm them with figures. EDGEBIC tracks how much of each work center's available capacity is consumed. A bottleneck sits near the top of that list and stays there across runs, often close to fully loaded while upstream machines run well below. If the lane looks full and the utilization figure agrees, the diagnosis is settled.

Three independent signals pointing at the same machine is as close to certainty as a diagnosis gets.

What a Bottleneck Is Not

Two lanes can look full without being constraints, and mistaking them for the real thing wastes effort. The first is the lane that is busy today and idle tomorrow. A machine can have a heavy afternoon without being the plant's pacing resource, so a single packed day proves nothing. Widen the horizon and watch whether the density holds; a constraint stays solid, a busy patch does not.

The second is the lane made full by one enormous job passing through, rather than by a genuine capacity shortfall. If a single long operation packs a lane this week but the machine is otherwise loose, the machine is not your constraint; that one job is heavy. Use the left-panel job filter to uncheck the big job and look again. If the lane opens up, the machine is fine and the job was the story. If the lane stays dense with the big job hidden, the machine is carrying broad load and the constraint diagnosis holds. Filtering is how you separate a busy machine from a bottleneck.

Telling the Engine to Respect It

Spotting the constraint is half the job. The other half is making the schedule work with it rather than against it. EDGEBIC lets you flag a work center as a bottleneck, which changes how the engine plans. Instead of pushing work forward and hoping the constraint keeps up, the engine anchors the schedule around the flagged machine and schedules backward from it, protecting its throughput so it is never starved and never the surprise that blows a due date.

That flag is a deliberate decision, separate from simply noticing a full lane. The two reinforce each other: you read the constraint off the Gantt, confirm it with utilization, then flag it so the plan is built around it. Setting the flag is a short task covered in how to flag and schedule around a bottleneck.

A Worked Example: The Oven Everyone Waits On

Acme Industries runs mills, an assembly cell, and one heat-treat oven. Deliveries keep slipping and nobody agrees why. A planner opens the Gantt.

The mill lanes have gaps: a bar, a pause, another bar. The assembly lane is similar. The heat-treat lane is a solid wall of bars from the first visible day to the last, with jobs queued right up against it. That is the first signal. The planner drags one heat-treat operation a day later to test a maintenance window and presses save. Fourteen bars across six jobs turn amber, because almost every job on the floor passes through the oven and everything after it now depends on the new timing. That is the second signal. The utilization list confirms the third: heat-treat runs near fully loaded while the mills sit lower.

The planner flags heat-treat as a bottleneck. On the next run the engine anchors the schedule around the oven, scheduling backward from it so parts arrive just as it is ready and its capacity is never wasted waiting. The mills, which were finishing early and stacking work in front of the oven, are re-timed to feed it at its pace. Nothing about the oven's speed changed. What changed is that the whole plan now bends around the real constraint instead of ignoring it, and the slipping deliveries that were caused by the oven being scheduled last, not first, come back under control.

The Constraint Sets the Pace

There is a long history behind this. GE Railcar took on-time shipping from 30 percent to 90 percent, and the mechanism was not more machines; it was scheduling the whole plant around its constraint. User Solutions has been building that discipline into scheduling tools since 1991, for shops and for names like Cummins and BAE Systems. The Gantt is where the discipline starts, because before you can schedule around a bottleneck you have to see it, and a full lane on a shared timeline is the plainest way to see it there is.

Find your fullest lane, confirm it, flag it, and let the plan respect it. Bring your real schedule to a demo and we will find the constraint together. Give US the lane that never has a gap.