Structures & MRP

Bills of Materials, Recipes and Kits

A structure says what something is made of. Manufacturing calls it a bill of materials, food production a recipe, distribution a kit — the same shape, modelled once.

Where to find it

Architect Panel → ERP - Operations:

• Structures — the console — build, explode and review
• Structure Records — the underlying parent/child lines
• Standard Costs — the costed view of a structure

One graph, any depth

A parent with children, where a child can itself be a parent. An assembly contains sub-assemblies containing components, to whatever depth your product genuinely has. There is no fixed number of levels.

What each line carries

• Quantity — how many of the child go into one parent.
• Scrap — expected loss, so requirements reflect what you must buy rather than what ends up in the product.
• Sequence — the order components are consumed or assembled, which drives works instructions.
• Revision — which version this line belongs to.
• Effectivity dates — when this line applies.

Effectivity is not optional

A structure without dates can only tell you what a product is made of now. The question actually asked is what it was made of in March — and it is asked in the worst circumstances: after a recall, a customer complaint or a supplier defect notice.

Change a structure by ending the old line's effectivity and starting a new one from the changeover date, so both answers remain available depending on the date you ask about.

Planning a changeover

Because effectivity is dated, a future change can be entered in advance. Set the new line to start on the changeover date and end the old one the day before, and the system builds correctly on the day without anybody making an edit at midnight.

Check remaining stock of the outgoing component when choosing the date — a changeover that strands inventory is an expensive way to be tidy.

Kits versus assemblies

The difference is what happens to stock. A kit is picked and shipped as its components and the parent may never exist as stock. An assembly consumes components and receives the parent into stock as a new item.

Decide which you are modelling before configuring — they produce different movements, and converting later means reworking the structure and the history.

Scrap is not a rounding fudge

Use scrap for genuine, predictable loss — offcuts, evaporation, test failures. Do not use it to compensate for inaccurate quantities, because it disguises the inaccuracy in exactly the reporting that would otherwise reveal it.

Explosions are dated too

Exploding for a works order due next month uses the structure effective then, not today's. If an explosion looks wrong, check the date you asked about before checking the structure.

Worked example

A manufacturer replaces a fastener from 1 September. The old line is ended 31 August, the new one starts 1 September, both entered in July. Builds through August use the old part; September builds use the new one. When a defect notice arrives in November covering the old fastener, exploding by date identifies exactly the builds that used it.

Recommendations

• Always set effectivity dates, even on a structure you think will never change.
• Enter changeovers in advance rather than on the day.
• Import structures deepest level first so a parent never references a missing child.
• Verify a sample explosion against your old system before trusting the set.

Demand, Supply and Planning

Planning takes demand, explodes it through the structures, nets off what you have and what is already coming, and says what must be made or bought and by when.

Where to find it

Architect Panel → ERP - Operations:

• Demand — what is required — orders, forecasts, minimum levels
• Supply — what is already coming — purchase orders, works orders
• Stock Policies — reorder levels and lead times per item
• Structures — the graph demand is exploded through

Netting

Requirements are not gross. Need 100, hold 30, with 20 on order, and the requirement is 50.

That is why stock accuracy determines whether planning output is useful at all. Planning against wrong stock produces wrong suggestions with complete confidence, and people act on them.

Lead times

Working backwards from a due date through each level's lead time produces the start dates. A component with a six-week lead time needs ordering six weeks before it is consumed, not before the finished item is due — which on a three-level structure can be a month apart.

Enter the lead times your suppliers actually achieve, not the ones they quote. The difference is usually the whole reason a plan slips.

Planning suggests, it does not act

Output is a set of suggestions for a person to review and convert into real orders. Nothing is placed automatically.

Treat a run producing surprising suggestions as a question about the data — usually stock, lead times or an effectivity date — rather than as an instruction. The first few runs after go-live are largely a data-quality exercise, and that is normal.

Before you rely on it

• Stock must be accurate, including reservations.
• Structures must be complete and correctly dated.
• Lead times must be realistic.
• Open purchase orders must be recorded as supply, or planning will suggest buying what is already coming.

Run planning in parallel with your existing method for a cycle or two and compare before anybody acts on the output.

Worked example

A works order for 500 assemblies is due in eight weeks. Planning explodes it: 500 housings held in stock, 1,000 fasteners needed against 200 held with 300 on order, so 500 to buy — and with a four-week lead time, ordered within two weeks or the build slips. The buyer sees one suggestion with a date rather than working it out from a spreadsheet.

Recommendations

• Fix stock accuracy before enabling planning. Everything downstream depends on it.
• Record open purchase orders as supply from day one.
• Use achieved lead times, and review them annually.
• Run in parallel for two cycles before anybody acts on suggestions.

Inspections and Non-Conformances

Where quality matters, receipts and production output are inspected before the stock is treated as good.

Where to find it

Architect Panel → ERP - Operations:

• Inspections — what was checked, by whom, and the result
• Non-Conformances — what failed, the disposition and the follow-up
• Stock Lots — the batch an inspection or failure attaches to

Inspections

An inspection records a check against a receipt, a lot or a production output — what was checked, who checked it, when, and the outcome.

Recording the inspector and the time is what makes the record evidence rather than an assertion. "It was inspected" is a claim; "inspected by J. Patel at 14:20 on the 3rd against lot 4471" is a record.

Non-conformances

When something fails, a non-conformance captures what was wrong, how much is affected, and what is to be done — use as is, rework, return to supplier, scrap.

The disposition matters as much as the failure, because it determines the stock movement that follows. A scrap disposition writes stock off; a return generates a movement back to the supplier; use-as-is leaves the stock but records that a decision was taken.

Tie it to lots

Record the lot on the inspection and on the non-conformance. That is what connects a quality failure to genealogy, so you can answer which outputs used the affected batch and where they went.

An NCR without a lot tells you something went wrong but not what to do about it, which is half a record.

Use it for supplier performance

Non-conformances aggregated by supplier are the most honest supplier quality metric you will have, because they are recorded at the point of failure by the person who found it — rather than compiled at review time by somebody reconstructing the quarter.

Analytics can report on them like any other datastore, so a supplier review can open with the actual numbers.

Keep the process light

An inspection regime nobody completes is worse than none, because it produces a record that looks like assurance and is not. Inspect what genuinely carries risk, make recording it quick, and resist adding checks after every incident until the form takes ten minutes.

Worked example

A goods-in inspection fails a delivery of seals on dimensional tolerance. The NCR records the lot, quantity affected and a return-to-supplier disposition, generating the movement. Genealogy confirms none of the lot reached production. At the quarterly supplier review, three NCRs against that supplier in six months are the agenda rather than an impression.

Recommendations

• Always record the lot.
• Record the disposition, not just the failure.
• Inspect by risk, not uniformly.
• Report NCRs by supplier and use them in reviews.