The EDGEBIC Setup Matrix: Sequence-Dependent Changeover, Explained

A sequence-dependent setup matrix replaces a single setup number on a routing step with a lookup of what the machine just ran and what it runs next. In EDGEBIC by User Solutions that lookup is defined per work center, entered in minutes, and organized in three tiers so a shop with a hundred products does not have to fill in ten thousand cells. The result is a schedule whose changeover hours match the ones the floor actually spends.

If you want the family concept defined on its own, read what is a setup family. If you want the arithmetic run across six paint jobs, read the paint-booth changeover walkthrough. This post is the capability: what it is, when you need it, and what it costs to maintain.

Why One Setup Number Is Not Enough

Ask a paint shop supervisor how long a changeover takes and the honest answer is a question: "from what to what?" White to cream is five minutes of purge. Black to white is four hours of solvent flush and color match. Red to white is three. Same booth, same crew, same day, a 48-fold difference in time.

A routing step that holds one setup number has to pick a single value for all of it. Every option is wrong:

• Pick the average and the light changeovers look expensive while the dark ones look free. The plan is wrong in both directions and the errors do not cancel, because sequence decides which errors you get.
• Pick the worst case and you refuse work you could have taken, because the schedule says the booth is full when it is not.
• Pick the best case, which is what most shops do, and the schedule promises deliveries the floor cannot make.

The third failure is the expensive one, because it looks like success until the orders are late. Three paint jobs sequenced white, black, white with a flat 30-minute setup finish on paper at 12:45. On the floor the two real flushes are 240 minutes each and the day runs about 11 hours. Nobody typed a wrong number. The data model simply could not hold the right one.

Changeover reduction has been a standard improvement target for decades and is one of the practices the Manufacturing Extension Partnership works on with small manufacturers. The prerequisite that gets skipped is measurement: you cannot schedule around a cost you have never written down.

The Three Tiers

EDGEBIC stores changeover in three layers that stack, not compete.

Families do the heavy lifting. A hundred products on one booth is a ten thousand cell problem stated product by product. Grouped into eight families it is a sixty-four cell problem, and most of those cells are zero. That reduction is the difference between a feature planners maintain and one they abandon after the first afternoon.

The family matrix is where you spend your time. It is a small grid: for each pair of families, how many minutes does the changeover take on this machine? Light to light is often zero. Light to dark might be 60. Dark to light might be 240. Fill the cells that matter and leave the rest.

Product overrides cover the exceptions. Red belongs to the light family, so red to black would normally take the light-to-dark time of 60 minutes. But red specifically needs an extra solvent flush, so a product-level cell says 90 minutes for exactly that pair. The override wins whenever it exists and the family rule covers everything else.

Why It Lives on the Work Center

The matrix is defined per work center, and that is deliberate rather than incidental.

Changeover physics belong to the machine. Two booths of the same model can have different flush times because one has older seals or a longer line. A furnace that ramps slowly needs more alloy-change time than the identical unit next to it. Storing the matrix per machine lets those differences be true at the same time.

It also makes routing decisions visible. If a product pair costs 240 minutes on booth one and 30 on booth two, the schedule now knows something useful about where that job should run. A matrix defined per product would flatten exactly the information you want.

Minutes In, Hours Out

Matrix values are entered in minutes. Planners say "a thirty-minute flush" and "a four-hour changeover", not "0.5 decimal hours", so the field takes the number they already have on the wall. The engine converts to hours internally.

Zero is a valid and useful value. A family-to-family cell of zero says these two groups run back to back with no changeover at all, which is precisely the fact you want the schedule to exploit when you sequence like with like.

What the Schedule Tells You Afterwards

Every scheduled operation records where its setup time came from. That audit trail is the part planners end up using daily, because it turns "why is that number 4 hours?" from an argument into a lookup.

Alongside the code sits a readable sentence, for example "Family matrix (Light to Dark) for White to Black: 60 min." The job view shows matrix-driven rows at full brightness and the fallback rows in a muted style, so you can scan a schedule and see immediately which changeovers the sequence logic actually priced.

That distinction is worth more than it sounds. A screen full of muted rows means your matrix is not covering the transitions that are happening, which is usually a missing family assignment rather than a missing cell. The five-level resolution order behind those codes is covered in how EDGEBIC resolves setup time.

Do You Need It?

You need a setup matrix when changeover time varies by more than a rounding error depending on sequence. In practice that means:

• Paint, coating, and printing. Color changes are the canonical case: light to dark is cheap, dark to light is expensive.
• Chemicals and food. Allergen changeovers, flavor purges, and cleaning validation between incompatible products.
• Heat treat. Alloy or temperature changes between recipes, with same-recipe runs costing nothing.
• Plastics and extrusion. Resin and color purges, where the purge quantity depends on both materials.
• Textiles. Dye lots and the same light-before-dark logic as paint.

You do not need it when your changeover is genuinely the same regardless of what ran before: a fixture swap that takes 20 minutes whatever the part, a tool change with a fixed cycle. Leave the flat routing setup in place. The matrix is not a general improvement, it is a specific answer to a specific kind of variation, and shops that do not have that variation gain nothing from maintaining one. For those cases the general practice in changeover time reduction is the better investment.

What It Costs, and What It Does Not Do

The maintenance cost is small and front-loaded. Group your products once, fill the cells that hurt, add the two or three overrides your team argues about. If a pair has no cell, the routing default applies and nothing breaks, so a partly filled matrix is a working matrix.

The runtime cost is zero. The matrix is loaded once before each scheduling run and every lookup is a direct table read. With no entries at all, the engine skips the mechanism entirely and behaves exactly as it did before, which means turning the feature on cannot change any schedule until somebody enters a cell.

It does not resequence your jobs for you. The matrix prices the changeovers in whatever order the schedule produces. Making the cost visible is the platform's job; deciding to run the lights before the darks is a planning decision, and the paint-booth walkthrough shows what that decision was worth on six real jobs: 630 minutes of changeover down to 90.

It does not reduce setup time. It reports it. The reduction still comes from the shop floor work described in job shop setup time reduction. What the matrix gives you is the number to aim at and the evidence that the aim was good.

Where to Go Next

How to build a setup matrix walks the families, the grid, the overrides, and the export. How EDGEBIC resolves setup time covers the five-level chain, cold starts, and what happens on a reschedule. Setup matrix mistakes covers the configurations that leave the matrix silently unused.

For where changeover sits among the platform's other scheduling mechanisms, see the complete guide to EDGEBIC. Bring your two worst changeovers and the machine they happen on to a demo of EDGEBIC and we will price them in a live schedule.