MRP (Material Requirements Planning)

MRP is the oldest idea in manufacturing software and still the most useful one. Given a set of orders you have promised to deliver, it works backwards: it explodes each order through its bill of materials to find gross requirements, subtracts free stock and quantities already on order, and then offsets the remaining shortfall by lead time so that each material is ordered on the day that gets it here just in time to be consumed. The output is a dated action list — buy this yarn on Tuesday, start this dyeing batch on the 14th.

The logic is simple enough to run on a spreadsheet, and thousands of mills do. What breaks the spreadsheet is not the arithmetic. It is that textile materials do not behave like the bolts and brackets MRP was designed for.

The four MRP inputs

An MRP run needs four things, and it is only as good as the worst of them. The demand schedule (confirmed orders plus a forecast for anything you intend to make to stock). The [bill of materials](/glossary/bom) for every style being made. Current inventory, including what is physically in the warehouse, what is committed to other orders, and what is already on a purchase order but not yet received. And the planning parameters for each material: lead time, minimum order quantity, order multiple, and safety stock.

Most failed MRP implementations fail on the third input. A factory whose stock records are 85 percent accurate will get an MRP output that is confidently, precisely wrong, and the planners will go back to their spreadsheet within a month. Inventory accuracy is the entry fee for MRP, which is why [barcode and RFID capture](/glossary/barcode-rfid) usually needs to come first.

Why textile MRP is genuinely harder

Standard MRP assumes materials are fungible: a bolt is a bolt, so if you have 500 in stock, any 500 will do. Almost nothing in a textile supply chain is fungible in that way, and every place where the assumption breaks is a place where a generic MRP run produces an answer that cannot be executed.

**Dye lots are not substitutable mid-order.** Stock of 4,000 metres of a navy jersey looks like coverage for a 3,600-metre order. If those 4,000 metres are split across three [dye lots](/glossary/dye-lot), it may be coverage for nothing at all, because shipping a garment order cut from three different lots invites a shade-variation rejection. Textile MRP has to net against available stock *within a single lot*, or at minimum flag when a requirement can only be met by mixing lots.

**Lead times are long, layered and unreliable.** A finished fabric requirement decomposes into greige fabric, which decomposes into yarn, which decomposes into fibre. Each stage carries its own lead time, and imported yarn or speciality fibre can run to several months. The planning horizon that MRP has to cover is therefore far longer than the order book, which means a large fraction of textile MRP is driven by forecast rather than confirmed demand — and forecast error at the yarn stage is expensive to unwind.

**Greige and finished stock are different planning objects.** Greige fabric is colour-agnostic and can serve any order for that construction. Finished fabric is committed to a shade. Buying greige early is a hedge; dyeing it early is a bet. An MRP model that collapses both into one "fabric" item cannot express that distinction, and so cannot advise on it.

**Consumption is a rate, not a count.** A shirt does not consume "1 fabric". It consumes a length or a weight, derived from a marker, plus a wastage allowance. Change the size ratio of the order and the fabric requirement changes even though the piece count does not.

A worked netting example

Take an order for 5,000 shirts, with fabric consumption of 1.6 metres per shirt and a cutting wastage allowance of 5 percent. Gross fabric requirement is 5,000 x 1.6 x 1.05 = 8,400 metres.

Two legitimate answers, differing by a factor of three, from the same data. Which one is correct depends on the buyer's shade tolerance, on whether the two lots fall within an acceptable shade band, and on whether the order can be cut in lot-segregated bundles. A generic MRP engine has no vocabulary for that question and will simply return 1,900 metres, which is how mills end up short of fabric with stock on the racks.

MRP, MPS and capacity

MRP is a materials calculation. It assumes infinite capacity: if the arithmetic says start dyeing 8,400 metres on the 14th, MRP says start dyeing on the 14th, regardless of whether every dye vat is full that week. That is not a bug so much as a division of labour — materials and capacity are separate problems — but it means MRP output has to be reconciled against [capacity planning](/glossary/capacity-planning) before it becomes a real plan. A schedule that is materially feasible and capacity-infeasible will simply fail later and more expensively.

The usual sequence is a master production schedule that is capacity-checked at a coarse level, an MRP run against that schedule, and then a finite-capacity scheduling pass on the individual machines. Skipping the middle step is common and produces the classic symptom: materials arrive on time and sit, because the machine that was supposed to consume them is still running last month's order.

What MRP does and does not fix

MRP will not tell you what to sell, will not improve your forecast, and will not survive bad inventory data. What it does, reliably, is remove the two failure modes that dominate a manually planned factory: ordering material too late to arrive in time, and ordering material you already have. Both are silent — the first surfaces as a delayed shipment weeks later, the second as yarn nobody can account for at stocktake.

Vastra ERP runs MRP against lot-level rather than SKU-level fabric and yarn stock, so netting respects [dye lot](/glossary/dye-lot) continuity, and it plans greige and finished fabric as separate stages with their own lead times. Requirements flow into [supply chain and procurement](/features/supply-chain) as dated purchase suggestions rather than an undated shortage list.