MRP Explained: The 1960s Algorithm Still Running Your Factory

Every night, in thousands of factories, a calculation runs that was designed before the personal computer existed. It answers one plain question: which materials do we need, how many, and when do we have to order them so they arrive in time? That calculation is MRP — material requirements planning.

If you have ever wondered why the system says "order 130" when you clearly need 400, or why a purchase suggestion shows up with a date that is already in the past, this article is for you.

Why MRP was invented

Before MRP, most factories managed parts the way a corner shop manages milk: when the shelf gets low, order more. That works for things customers buy steadily. It works badly for parts that are needed only because something else is being built.

In the 1960s, Joseph Orlicky — an engineer who went on to champion the method at IBM — pointed out that a component's demand does not have to be guessed. It can be calculated. If you are building 200 pumps next month and each pump has two seals, you need 400 seals. No forecast required.

The only obstacle was arithmetic: thousands of parts, several levels of assemblies, different lead times for each. Computers had just become able to do that arithmetic overnight, and MRP was born. Orlicky's 1975 book turned it into an industry standard, and the same logic still sits at the heart of almost every ERP system today.

What MRP does, in three steps

1. Explode. Start from what you plan to build — the master production schedule — and walk down the bill of materials. 200 pumps × 2 seals = 400 seals. This is the gross requirement. Demand calculated this way is called dependent demand, and it is the reason you should never forecast screws.

2. Net. Subtract what you already have or already expect: stock on hand, plus open orders that will arrive before the need date. What remains is the net requirement.

3. Offset. Count back by the lead time. If seals take three weeks to arrive from the supplier, the order must go out three weeks before the seals are needed.

Then MRP repeats the same three steps one level down — for the components of anything you make in-house — until it reaches purchased materials.

A worked example

You need 400 seals in week 4. You have 120 on the shelf, and a supplier order for 150 is due in week 3. Lead time is three weeks.

You will have 270 seals against a need of 400, so you are 130 short. With a three-week lead time, the order for 130 must be placed in week 1.

Now suppose today is already week 2. MRP will show the same order with a release date in the past. That is not a bug. It is MRP telling you, politely, that the plan cannot be met as it stands: you need to expedite, find a faster supplier, or move the pump build.

Real systems add one more step, lot sizing. If seals come in boxes of 100, the 130 becomes 200, and the extra 70 show up as stock for later weeks.

The assumption everyone forgets

MRP assumes infinite capacity. It will happily plan 400 seals, 200 housings and 300 assembly hours into the same week without asking whether the assembly line has 300 hours to give. It checks materials, not machines.

That is why MRP never works alone. Rough-cut capacity planning (RCCP) tests the plan against key resources before materials are ordered, and production scheduling turns the result into a sequence that actually fits the machines. The story of how the industry bolted capacity onto MRP — and called it MRP II — is told in MRP vs MRP II.

Three myths about MRP

"If MRP is wrong, run it more often." MRP is only as good as its three inputs: the bill of materials, the stock balances and the open orders. If a BOM lists the wrong quantity, a goods receipt was booked two days late, or a purchase order carries a stale date, running MRP every hour simply produces a wrong answer every hour. Fix the data first.

"MRP tells buyers what to buy." It tells them what is missing. A shortage does not always need a new purchase order. Often the better answer is to pull an existing order forward, transfer stock from another site or change the production sequence — see six shortages that don't need a new purchase order.

"With MRP you don't need safety stock." MRP calculates dependent demand precisely, but it trusts lead times to be true. For parts with unreliable suppliers, or parts consumed steadily across many products, a buffer still makes sense. Safety stock and reorder points explain when.

Where MRP shines — and where it struggles

MRP is at its best in assembly-heavy production with clear bills of materials: pumps, electrical equipment, furniture, made-to-order cable. It struggles in three situations:

• Demand swings a lot. Every change ripples down every BOM level, and purchase suggestions flip from week to week. Planners call this nervousness.
• Lead times are long and unreliable. The offset step is only as good as the lead time behind it.
• The real limit is a machine, not a material. MRP will plan the materials perfectly for a week the bottleneck cannot deliver.

For the first two, many companies place deliberate stock buffers at a few chosen points in the product structure. That is the idea behind DDMRP.

How it looks in factory.online

In factory.online, MRP works from planned manufacturing orders through multi-level BOMs, stock and open supply, and keeps each shortage linked to the order that created it. A planner sees that the 130 seals belong to the pump order for a specific customer — not just an anonymous line on a purchase list. Capacity checks and scheduling run on the same data, so the infinite-capacity gap is closed in the same place instead of in a separate spreadsheet.

If your planners spend Monday mornings reconciling MRP output with Excel, that is a good place to start a conversation. Book a demo and bring one of those Monday lists with you.