Material requirements planning: how MRP works, step by step
Material requirements planning explained: the four steps from build plan to purchase order, how to trace any MRP suggestion, and when the plan breaks.
Material requirements planning (MRP) turns your build plan into purchase and work orders. It answers three questions for every part: what you need, how many, and when. It explodes the bill of materials, nets the result against stock and open orders, sizes each order, and offsets it by the lead time.
That chain explains every suggestion your ERP makes, including a purchase order for 600 brackets when the build needs 48, or an order that was due last week. Once you can run it by hand, you can check any suggestion and find the input that sent it wrong.
What is material requirements planning?
MRP reads three inputs: the master production schedule, which says which finished products you build in which week; the bill of materials, which says which parts each product uses and how many; and inventory records, which hold stock on hand and on order, plus each part's lead time, lot rule and safety stock.
MRP gives back planned orders, each with a quantity and a release date, and exception messages for existing orders that no longer fit the plan.
The MRP calculator runs the same grid in your browser, so you can change any number below and watch the plan move.
Try it: one bracket, two pumps
Take an illustrative purchased part, mounting bracket BR-210. Pump P-100 uses two and pump P-200 uses one. You have 420 on hand, an open PO for 500 due in week 2, a safety stock of 150, a 3-week lead time, 2% scrap, and a supplier minimum of 600 in multiples of 200. MRP builds this grid from those numbers:
Read it column by column. In week 1 the pumps need 120 brackets, 123 with scrap, and stock covers them. In week 2 the open PO lands and covers the week. In week 4 stock would fall to 102, 48 under safety stock, so MRP plans an order. The supplier's minimum turns 48 into 600, and the lead time moves the release back to week 1. Weeks 7 and 9 repeat the pattern.
Your buyer acts on the bottom row: place 600 now, 600 in week 4 and 600 in week 6.
How does MRP work, step by step?
MRP runs four steps for every part, one week at a time: explode, net, lot size, offset. Each step is a line of arithmetic you can check by hand. A wrong input at any step carries through to the order at the end, so the inputs deserve as much of your attention as the arithmetic.
Exploding the bill of materials
The gross requirement is everything the build plan consumes in a week. For a part used in several products, multiply each parent's builds by its quantity per, add them up, then add the scrap allowance:
Gross = (builds of P-100 × 2 + builds of P-200 × 1) × (1 + scrap %)
In week 4 that is (80 × 2 + 120 × 1) × 1.02 = 285.6, which MRP rounds up to 286. Planners lose sight of demand on shared components: neither pump's schedule looks big, yet together they need 357 brackets in week 6. If the floor issues 2.1 brackets per P-100, the bill of materials is wrong and every week runs short.
Some systems treat scrap as a yield and divide by (1 − scrap %). At 2% the two methods differ by less than half a unit in a thousand. At 20%, one asks for 1.20 parts per good part and the other for 1.25, so check which one your ERP uses.
Netting against stock and open orders
Netting subtracts what you have from what you need, and keeps safety stock intact:
Available = on hand at the end of last week + scheduled receipts − gross
Net = max(0, safety stock − available)
Written the long way, that is gross + safety stock − on hand − scheduled receipts, floored at zero. In week 4, week 3 leaves 388 and no open PO lands, so available is 388 − 286 = 102. Safety stock is 150, so the net requirement is 48.
Scheduled receipts are your open POs. MRP counts each one in full in the week it is due, and it cannot tell whether the supplier confirmed that date. If MRP suggests a PO for a part you have on order, check the open PO's due date first. When that PO lands after the need, MRP plans a new order to cover the gap. If the safety stock calculator gives you a different number than your item master holds, this step is where the difference shows up.
Offsetting by the lead time
You have to order a week 4 requirement before week 4. MRP subtracts the lead time from the need date:
Release week = need week − lead time
Week 4 minus 3 weeks of lead time is week 1. When the result falls before today, MRP still shows the order and marks it past due, so you can see the shortage coming. That is how MRP ends up telling you to place an order that is late before you place it.
Lot sizing and MOQ
The lot rule turns the net requirement into an order quantity. You will meet these three most:
- Lot-for-lot orders the net requirement: 48.
- Fixed quantity orders in whole lots of a set size: with a lot of 500, round 48 up to 500.
- MOQ with an order multiple orders at least the minimum, in steps of the multiple:
max(net, MOQ)rounded up to the next multiple, so 48 becomes 600.
The extra 552 go into projected on hand and cover the next few weeks, so the grid's next planned order comes in week 7. The same demand under each rule:
| Lot rule (illustrative) | Orders in 10 weeks | First order (need 48) | Units bought | Average on hand | On hand after week 10 |
|---|---|---|---|---|---|
| Lot-for-lot | 7 | 48 | 1,611 | 229 | 150 |
| Fixed quantity 500 | 4 | 500 | 2,000 | 465 | 539 |
| MOQ 600, multiples of 200 | 3 | 600 | 1,800 | 445 | 339 |
Lot-for-lot keeps stock lowest and places the most POs. Over ten weeks the MOQ buys 189 more brackets than lot-for-lot and holds about twice the average stock. The price break and your cost to hold stock decide if that trade pays. The EOQ calculator works it out, and MOQ vs EOQ covers when the minimum is worth paying.
Why did your ERP suggest that PO?
Trace it backward from the planned order's due date and walk the four steps in reverse. Find the week that needs the parts, the gross requirement and the parents behind it, what stock and open orders leave, how far under safety stock that falls, and the lot rule that raised the quantity.
Most ERPs show the demand behind each order. Planners call this pegging: each planned order links to the sales orders, jobs or forecast lines it covers. Check the pegging before you change anything. A suggestion that looks wrong often traces back to a forecast nobody removed or a job with the wrong quantity.
When MRP compares the new plan with your existing orders, it raises exception messages, also called action messages:
- Expedite or reschedule in: an open order is due after the week that needs it. Ask the supplier to ship sooner.
- Push out or reschedule out: an open order is due before you need it. Move the date out if the supplier can hold it.
- Cancel: the demand behind an open order has gone. Cancel or reduce the order.
- Release past due: you should have placed the planned order by now.
Unread expedite messages cost money, because each one marks a future shortage. Your ERP shows these messages next to its planned orders. Work the expedite messages first, then the cancels, which free cash.
Why does one short part stop the build?
MRP plans each part on its own grid, and the floor builds from kits. A pump starts only when every part in its kit is there, so a bracket at 98% and a bearing at 98% do not make a 98% build. The chance of a complete kit is the product of every part's chance.
If each of 10 purchased parts is on hand for 98% of builds, and the parts run short independently, the kit is complete 0.98 to the power of 10 = 81.7% of the time. About one build in five waits.
How often a whole kit is complete
An illustrative table of kit availability: the chance every purchased part in a kit is on hand when the build starts, for kits of 1, 5, 10 and 20 parts with each part on hand 99%, 98% or 95% of the time. A 10-part kit at 98% per part is complete 81.7% of the time; a 20-part kit at 95% per part is complete 35.8% of the time.
| Purchased parts in the kit | Each part 99% | Each part 98% | Each part 95% |
|---|---|---|---|
| 1 | 99.0% | 98.0% | 95.0% |
| 5 | 95.1% | 90.4% | 77.4% |
| 10 | 90.4% | 81.7% | 59.9% |
| 20 | 81.8% | 66.8% | 35.8% |
Each column multiplies one part's availability by itself once per part in the kit. Shortages are assumed to be independent.
Set component targets from the kit instead of part by part. Pick the kit availability the build needs, then take the root: a 95% kit target across 10 parts asks for 95% to the power of 1 ÷ 10 = 99.5% on each part. A part figure set this way is a cycle service level, the chance of no stockout before a build. Label it that way, because a fill rate, the share of demand served from stock, measures something else.
For components with a long cumulative lead time, use statistical safety stock sized to that kit floor. The MRP calculator works the kit arithmetic for any number of parts.
What happens when the typed lead time is wrong?
MRP treats the lead time in the item master as known and fixed. It offsets every order by that number and never compares it with how long your supplier takes. When the real lead time runs longer, every planned order arrives late, and MRP gives you no warning until the shortage shows up.
What a late supplier does to the plan
Keep the bracket plan as MRP made it with a 3-week lead time, and let the supplier take 5:
Typed lead time 3 weeks, actual 5
An illustrative comparison of projected on hand for the bracket when MRP plans with a 3-week lead time and the supplier takes 5 weeks. In MRP's plan stock never drops below the 150 safety stock. In reality each planned order lands two weeks late: stock falls to 102 in week 4, runs out at minus 62 in week 5, recovers, and is short again in weeks 7, 8 and 10, by up to 380 units in week 8.
| Row | Wk 1 | Wk 2 | Wk 3 | Wk 4 | Wk 5 | Wk 6 | Wk 7 | Wk 8 | Wk 9 | Wk 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| On hand, MRP's plan (3-week lead time) | 297 | 552 | 388 | 702 | 538 | 181 | 577 | 220 | 656 | 339 |
| On hand, supplier takes 5 weeks | 297 | 552 | 388 | 102 | -62 | 181 | -23 | -380 | 56 | -261 |
MRP kept the 3-week plan while the supplier took 5. Negative numbers are builds that wait for parts.
The order MRP planned for week 4 lands in week 6. Stock drops under safety stock in week 4, runs out in week 5, and runs short again in weeks 7, 8 and 10. With 5 weeks in the item master, MRP would have marked the first order past due on day one and told you to act. With 3 weeks typed in, the grid showed no problem until the parts failed to arrive.
Planners tend to expedite each shortage on its own, then raise safety stock to cover the next one. That ties up cash in every part from that supplier, including the parts that arrive on time.
Safety lead time or safety stock
When receipts land late and demand holds steady, the risk is timing, and safety lead time handles timing better than safety stock. Safety lead time plans each order to arrive a set number of weeks before the build needs it. Give the bracket 2 weeks of safety lead time and let the supplier take 5: stock never runs short. The cost shows up as orders arriving two weeks early, and the first order is past due today. Safety stock pays for the same slip with extra units held in every week. Short shipments are a separate risk, a yield problem that lead-time spread does not capture.
What to put in the lead time field
Enter the time from placing the PO to having the part in stock, ready to issue. Measure it from your own receipts, and use the supplier's quote only when you have no history. Then pick which number from those receipts to use:
- The median keeps stock low and runs late about half the time.
- A high percentile, such as the 90th, runs late about one time in ten. MRP orders earlier, so you hold more stock.
Use the high percentile for parts that can stop a build, and the median for parts with slack or cheap substitutes. Recheck the field every quarter. Your receipts also give you the spread: received date minus order date, per supplier and part. The safety stock formula SS = z × √(L × σd² + d² × σL²) takes that spread as σL, so a supplier who delivers on the same day every time earns a smaller buffer at the same target. A spread above about a quarter of the mean lead time marks unreliable supply. The supplier lead time lesson shows how to pull actual lead times and what a quote leaves out.
Keep strategic buffers for a single-source part, or for a disruption your receipt history has never seen, in their own layer above safety stock. Size them in days of cover, so the time you can run without the supplier is at least the time the supplier needs to recover. A separate layer shows what resilience costs.
Cartograph data: to be published after review. How far real supplier lead times move against the typed ones, measured from purchase orders in Cartograph. Planned fields: median and 90th-percentile actual lead time against planned lead time by part category, share of PO lines that arrive after the promised date, median slip in days for late lines, and share of open PO lines with no supplier confirmation.
MRP plans with the dates you give it. Cartograph follows up on open POs, and Scout Agent emails and calls your suppliers to confirm dates and expedite late lines. When a supplier's new date means a receipt will land after the build needs it, a readiness signal opens a task, and you hear about it before the line runs short.
When is the MRP math wrong?
MRP's arithmetic holds up. Bad inputs send the plan wrong: a stale bill of materials, on-hand counts that are off, unconfirmed open POs, a schedule the plant cannot build, an MOQ that overbuys, or a part that never needed MRP. Each one fails in its own way, so check them on the parts that gate your builds first.
What each one does:
- Stale bill of materials. A wrong quantity per or a missing scrap allowance multiplies through every week.
- Wrong on hand. Unposted receipts, stock in quarantine, or a second warehouse counted twice change every net requirement.
- Unconfirmed open POs. MRP counts every open PO in full on its due date. A due date the supplier never confirmed is a guess. See purchase order tracking.
- Infinite capacity. MRP assumes your plant can build the whole master schedule. If the line cannot run a week's builds, the parts for that week arrive early and sit on the shelf.
- MOQ overbuy. Turning a 48-unit need into a 600-unit order is correct arithmetic and may still be the wrong buy. Price the leftover before you accept it.
- The wrong tool for the part. For cheap parts with steady use, a reorder point or a two-bin kanban does the job with less setup than MRP and the same protection. The reorder point formula sets one.
A plan can pass clear-to-build on paper and still fail on the floor if any of these is off. Check the inputs on the parts that gate your builds first.
What is the difference between MRP, MRP II and ERP?
MRP plans materials: what to buy or make, and when. MRP II, manufacturing resource planning, adds capacity, shop floor control and the financial side of the same plan. ERP, enterprise resource planning, wraps all of that together with sales, purchasing and accounting in one system. Most ERPs now carry MRP as one module.
Questions teams ask
What is the difference between gross and net requirements?
The gross requirement is everything the build plan consumes in a week: builds times quantity per, plus scrap. The net requirement is what you still need after stock and open orders cover what they can, with safety stock kept intact. MRP plans orders for net requirements only.
How does an MOQ change an MRP suggestion?
MRP raises the order to the minimum, then rounds up to the order multiple. The extra units go into projected on hand and cover later weeks, which pushes the next order out.
How does MRP calculate when to order?
It finds the first week projected on hand would fall below safety stock, plans an order to arrive that week, and subtracts the lead time to get the week you place the PO. A need in week 4 with a 3-week lead time puts the release in week 1. If the result falls before this week, the order is past due.
Why did my ERP suggest this purchase order quantity?
Start with the week the order is due. Take the gross requirement, subtract what stock and open POs leave, and add back safety stock: that is the net requirement. Then apply the lot rule. When a suggestion runs far above the need, look at the minimum order quantity, the order multiple or a fixed lot size.
Does MRP use safety stock?
Most systems can. In the bracket example safety stock works as a floor: MRP plans an order whenever projected on hand would drop below it. Some ERPs leave safety stock out of netting and only report it, so check how yours is set.
Why does MRP suggest an order that is already late?
The need date minus the lead time falls before today. MRP still shows the order so you can see the shortage. Place it, ask the supplier if they can expedite, and check whether the build can move.
What is the difference between MRP and a reorder point?
A reorder point triggers an order when stock falls to a level set from average use. MRP orders for specific future builds, so it handles lumpy demand from a production schedule. Many plants run MRP for parts on the bill of materials and reorder points for cheap, steady consumables.
What lead time should I enter in my ERP?
The time from placing the PO to having the part in stock, ready to issue, measured from your own receipts. For parts that can stop a build, use a slow delivery from that history, such as the 90th percentile. Recheck it every quarter.
What are the benefits of MRP?
MRP orders each part for the week the build needs it, so you hold less stock than with a fixed reorder level on every part. It shows shortages weeks ahead, flags open orders that no longer fit the plan, and ties every suggestion back to a build you can check.
Can I run MRP in Excel?
Yes, for a handful of parts. Put weeks across the columns and one row each for builds, gross requirements, open POs, projected on hand, net requirements and planned orders. The MRP calculator has a free Excel template with the formulas set up. Past a few dozen parts with shared components, the run belongs in your ERP.
How should I set service levels for parts in a kit?
Start from the kit. Pick how often the build must start with every part on hand, then give each part that target to the power of 1 ÷ the number of parts. A 95% kit target across 10 parts asks for 99.5% per part. Cheap parts can carry more than that, which lets an expensive part run a little lower while the kit still hits its target.
Does every ERP calculate MRP the same way?
Every MRP run follows the same chain: explode, net, lot size and offset. The settings differ: how scrap is applied, whether safety stock is netted or only reported, whether lead time is in days or weeks, and which lot rules you can pick. Check those settings before you compare a suggestion with the bracket grid above.
If your team spends hours each week working out which open POs will land after the build needs them, talk to our team about how Cartograph confirms those dates with your suppliers.