Chapter 5: Material Requirements Planning - The Classic Brain |
5.1 The Question That Changed Manufacturing |
In the 1960s, a quiet revolution began in American factories. A small group of production planners and computer pioneers asked a question that seems obvious in hindsight but was radical at the time: What if we stopped ordering materials based on guesses and instead calculated exactly what we need, exactly when we need it, using the actual production schedule |
Before that question was answered, manufacturing was a world of hunches. A planner would look at a shelf of steel bars and think, 'We are running low. I should order more.' They would call a supplier and order a quantity that felt right - perhaps a full truckload, because that was simpler. The result was either too much material, sitting and rusting, or too little, stopping production. There was no science. There was only experience and intuition. |
The answer to that question became Material Requirements Planning (MRP) . MRP was the first truly algorithmic approach to manufacturing planning. It was the classic brain of the factory - the logical engine that takes a production schedule and transforms it into a precise set of purchase orders and work orders. Even today, after decades of evolution, MRP remains the core intellectual engine of every ERP system for mechanical manufacturing. Everything else - inventory tracking, capacity planning, shop floor control - serves and supports MRP. |
This chapter explains what MRP is, how it thinks, why it is so powerful, and where it has limits. You will learn why MRP is often called the 'classic brain' and why, despite its age, it is still the most important planning tool for any factory that makes complex products from many components. |
|
5.2 The Pre-MRP Darkness - A Glimpse into the Past |
To appreciate MRP, you must first understand the chaos that existed before it. Imagine a mechanical factory in the 1950s. The factory makes gearboxes. Each gearbox contains dozens of parts: cast housings, machined gears, shafts, bearings, seals, and fasteners. The factory has a master schedule - a handwritten list of what gearboxes to build each week. But there is no systematic way to translate that schedule into material orders. |
The planner sits at a desk covered in paper. For each part, they look at the master schedule, look at the bill of materials, and manually calculate how many they will need. They subtract what is already in stock and what is already on order. Then they place purchase orders for the difference. This calculation is repeated for every part, every week. It is tedious, error-prone, and slow. |
The real problem is not the tedium. The real problem is timing. A gearbox housing is made from a casting. The casting takes four weeks from order to delivery. The planner must order the casting four weeks before the gearbox is scheduled for assembly. But the planner has no systematic way to offset lead times. They might remember that the casting has a long lead time and order early. But what about the sealsThey have a short lead time. Order them too early, and they sit in inventory. Order them too late, and production stops. |
Every week, the planner makes hundreds of decisions. Many are wrong. Some are wrong in the direction of excess inventory. Some are wrong in the direction of shortages. The planner cannot know which is which until weeks later, when the materials either arrive or do not. By then, it is too late to fix. |
This was the world before MRP. It worked, after a fashion, because factories were simpler, products were less complex, and customers were more patient. But as mechanical products became more complex and competition intensified, the old ways stopped working. |
|
5.3 The Core Logic - How MRP Thinks |
MRP replaces guesswork with a logical procedure. The procedure is not magic. It is simply a disciplined way to answer four questions for every part that goes into a finished product. The first question is: What do we needThe second question is: How many do we needThe third question is: When do we need themThe fourth question is: What do we already have |
To answer these questions, MRP uses three inputs. The first input is the Master Production Schedule (MPS) , which we explored in Chapter 4. The MPS tells MRP what finished products are needed and when. The second input is the Bill of Materials (BOM) , which we explored in Chapter 2. The BOM tells MRP what components go into each finished product and in what quantities. The third input is inventory status , which tells MRP what materials are already on hand, what is already on order from suppliers, and what is already in production. |
MRP processes these inputs through a logical operation called BOM explosion. Explosion starts at the top level - the finished product. For each week in the MPS, MRP multiplies the quantity of finished products by the quantities in the BOM. This yields the gross requirements for components at the next level down. Then MRP subtracts existing inventory and scheduled receipts to yield net requirements. Then MRP offsets those net requirements by the lead times of each component, moving the need date earlier. Then MRP explodes down to the next level, treating each component as a parent for its own children. This repeats until every level of the BOM is processed. |
What makes MRP powerful is that it respects the dependent demand nature of components. The demand for a gear is not independent - it does not come from customers directly. It depends on the demand for the gearbox that contains the gear. If you need ten gearboxes, you need ten gears. If you need zero gearboxes, you need zero gears. This seems obvious. But before MRP, many factories planned each component independently, as if the demand for gears and the demand for gearboxes were unrelated. The result was chaos. |
MRP also respects lead time offsets. If you need a casting in week ten to assemble a gearbox in week eleven, and the casting has a four-week lead time, MRP creates a planned order release for the casting in week six. The system calculates this automatically. The planner does not need to remember that castings take four weeks. The ERP stores the lead time for each part, and MRP uses it. |
|
5.4 The Explosion - A Walk Through the Logic |
Let us walk through a simple explosion example. We will keep the numbers small and avoid formulas, but the logic will be clear. Imagine a mechanical factory that makes a small hydraulic pump. The pump has three levels. At the top level is the finished pump. At the second level are the pump body, the piston assembly, and the seal kit. At the third level are the raw materials for the pump body and the components of the piston assembly. |
The MPS says: build fifty pumps in week twelve. MRP starts with the top level. The gross requirement for pumps in week twelve is fifty. There are no pumps in inventory, so the net requirement is also fifty. The lead time to assemble a pump is one week. So MRP creates a planned order release for fifty pumps in week eleven - meaning that assembly should start in week eleven to finish in week twelve. |
Now MRP explodes to the second level. For each pump, the BOM says: one pump body, one piston assembly, and one seal kit. So the gross requirements for these components in week eleven are fifty each. Why week elevenBecause the components must be available when assembly starts. |
MRP checks inventory for each component. There are ten pump bodies in stock, so the net requirement for pump bodies is forty. There are five piston assemblies in stock, so the net requirement is forty-five. There are zero seal kits in stock, so the net requirement is fifty. |
Now MRP applies lead times. The pump body is machined from a casting. The machining lead time is two weeks. So the planned order release for pump bodies is week nine - two weeks before they are needed in week eleven. The piston assembly is built from components. Its lead time is one week. So the planned order release for piston assemblies is week ten. The seal kit is purchased. Its supplier lead time is three weeks. So the planned order release for seal kits is week eight. |
Now MRP explodes to the third level. For the pump body, the BOM says: one casting. The gross requirement for castings is forty in week nine. Inventory shows fifteen castings in stock, so the net requirement is twenty-five. The casting supplier has a four-week lead time. So MRP creates a planned purchase order release for twenty-five castings in week five. |
For the piston assembly, the BOM says: one piston rod and one piston head. The gross requirements are forty-five each in week ten. Inventory shows zero piston rods and ten piston heads. So the net requirement is forty-five piston rods and thirty-five piston heads. The piston rod is made from steel bar. The machining lead time is one week, and the steel bar has a two-week purchasing lead time. So MRP creates a planned purchase order for steel bar in week seven and a planned work order for the rods in week nine. The piston head is purchased as a finished part with a two-week lead time. So MRP creates a planned purchase order for thirty-five piston heads in week eight. |
By the end of this explosion, MRP has generated a complete set of planned orders for every component and raw material, with due dates that are perfectly coordinated. The castings will arrive in week nine, just when they are needed to make pump bodies. The pump bodies will be finished in week eleven, just when they are needed for assembly. The seal kits will arrive in week eleven, directly from the supplier to assembly. Everything flows together like a well-choreographed dance. |
This is the magic of MRP. It is not magic. It is simple arithmetic, repeated thousands of times, executed by a computer that never gets tired and never forgets to offset a lead time. |
|
5.5 The Logic of Netting - What You Have Versus What You Need |
One of the most valuable aspects of MRP is netting - the process of subtracting what you already have from what you need. Netting sounds trivial, but in a real factory, it is surprisingly complex. MRP must consider not only the inventory on hand but also what is already on order from suppliers and what is already in production. |
Imagine a simple example. You need one hundred steel bars in week ten. You have twenty bars in inventory. You already placed a purchase order for fifty bars that is scheduled to arrive in week nine. You have a work order for thirty bars that will be produced internally in week eight. MRP does not simply subtract the twenty on-hand bars from the one hundred needed. It also subtracts the fifty coming from the supplier and the thirty coming from internal production. The net requirement is zero. You need no new orders. |
But what if the timing does not matchWhat if the fifty bars from the supplier are scheduled to arrive in week eleven, but you need them in week tenThen MRP cannot use them. They arrive too late. The net requirement becomes one hundred minus twenty on-hand minus zero usable from the supplier - so eighty bars. MRP will then create a new planned order for eighty bars with a due date of week ten. |
This is called time-phased netting. It is the heart of MRP's intelligence. A simple subtraction of totals would have told you that you have one hundred bars in total (twenty on-hand, fifty on order, thirty in production), which exactly meets the need for one hundred. But because the timing is wrong, you are actually short. MRP catches this timing problem automatically. A human planner, overwhelmed by hundreds of parts, would likely miss it. |
Time-phased netting is the reason MRP is so effective at preventing shortages. It does not just ask 'How many' It asks 'How many, and when' The difference is everything. |
|
5.6 Low-Level Coding - Solving the Multiple-Use Problem |
A mechanical part is often used in many different products. The same steel bar might be used in pump bodies, piston rods, and valve housings. The same bearing might appear in a gearbox, a conveyor roller, and a fan assembly. This creates a challenge for MRP. If you calculate requirements for the steel bar separately for each product, you might count the same bar multiple times. Or you might miss the fact that a single bar can serve multiple purposes. |
MRP solves this with a technique called low-level coding. Every part in the BOM is assigned a code that represents the lowest level at which it appears in any BOM. A part that appears only at level three of a pump BOM but never anywhere else has a low-level code of three. A part that appears at level two of a gearbox BOM and level three of a pump BOM has a low-level code of three, because three is lower than two. |
Why does low-level coding matterBecause MRP processes the BOM level by level, from the top down. It calculates gross requirements for all parts at level zero, then level one, then level two, then level three. When MRP reaches a part's low-level code, it has already collected all gross requirements for that part from all higher levels. It can then net everything at once, rather than netting partially for each product and risking double-counting or under-counting. |
Low-level coding is invisible to the user. The ERP system handles it automatically. But it is a critical piece of MRP's logical foundation. Without it, MRP would be correct only for products that do not share components. In mechanical manufacturing, where component sharing is the norm, low-level coding is essential. |
|
5.7 The Role of Safety Stock - MRP Is Not a Fortune Teller |
MRP assumes that the future is known. It assumes that the MPS is accurate, that lead times are fixed, that suppliers deliver on time, and that the factory runs without disruptions. These assumptions are never true. The real world is full of surprises. A customer might increase an order at the last minute. A shipment might be delayed by a port strike. A machine might break down, causing production to fall behind. |
MRP does not ignore these uncertainties. Instead, it provides a tool to handle them: safety stock. Safety stock is extra inventory held beyond the calculated requirements, as a buffer against uncertainty. If demand is higher than forecast, safety stock absorbs the difference. If a delivery is late, safety stock covers the gap until the materials arrive. |
Safety stock is not a guess. The ERP calculates safety stock levels based on historical variability. For a component with highly variable demand, the ERP suggests a larger safety stock. For a component with a very reliable supplier, the ERP suggests a smaller safety stock. For a component that is cheap and small, safety stock is generous. For a component that is expensive and large, safety stock is tight. |
But safety stock is a compromise. It reduces the risk of shortages, but it increases inventory holding costs. The ERP does not decide the trade-off. The factory management decides, based on their priorities. A factory that serves aerospace customers with high penalties for late delivery might choose generous safety stock. A factory that makes low-margin commodity parts might choose minimal safety stock, accepting some risk of shortage in exchange for lower inventory costs. |
The important point is that safety stock is integrated into MRP. When MRP calculates net requirements, it subtracts not only on-hand inventory but also the designated safety stock. The safety stock is reserved for emergencies; it is not available for routine consumption. If a planner tries to use the safety stock for a normal order, the ERP will show a negative available quantity, forcing a conscious decision to consume the buffer. |
|
5.8 The Action Messages - What the Planner Actually Does |
A full MRP run for a medium-sized mechanical factory might generate tens of thousands of planned orders. No planner can review each one individually. The ERP system helps by creating action messages - short, prioritized instructions that tell the planner what needs attention. |
Action messages fall into several categories. A release message tells the planner that a planned order is now firm enough to be converted into a real purchase order or work order. The timing has arrived, the materials are available, and the capacity is free. The planner clicks a button, and the ERP creates the actual order. |
An expedite message tells the planner that a previously released order is at risk of being late. The due date is approaching, and the order is not yet complete. The planner must investigate: call the supplier, check the shop floor, or escalate to management. |
A de-expedite message tells the planner that a previously released order is no longer needed as early as planned. Perhaps a customer order was delayed, or the factory built ahead of schedule. The planner can postpone the order, saving inventory carrying costs or freeing capacity for more urgent work. |
A cancel message tells the planner that a planned order is no longer needed at all. Perhaps the customer canceled the order, or the BOM changed, or inventory unexpectedly appeared. The planner cancels the order to avoid unnecessary procurement or production. |
A reschedule message tells the planner that an order's due date needs to move, but not so urgently as expedite or de-expedite. The planner adjusts the date in the system, and the ERP propagates the change to dependent orders. |
Action messages are prioritized by urgency and impact. A message about a part that is needed tomorrow for a customer order that is already late appears in red, at the top of the list. A message about a part that is needed in six weeks, with plenty of inventory, appears in green, at the bottom. The planner works the list from top to bottom, addressing the most critical issues first. |
Without action messages, MRP would be an overwhelming flood of data. With action messages, MRP becomes a manageable, focused to-do list for the planner. This is why MRP is called a planning tool, not an automation tool. It does not replace the planner. It tells the planner where to look. |
|
5.9 Closed-Loop MRP - Adding Feedback |
The original MRP systems of the 1960s and 1970s were open-loop. They calculated requirements and generated orders, but they did not check whether those orders could actually be executed given capacity constraints. A planner might follow MRP perfectly, only to discover that the required work overloaded a critical machine. The plan was mathematically correct but practically impossible. |
This led to closed-loop MRP. Closed-loop MRP adds two feedback mechanisms. The first is capacity feedback. After MRP generates planned orders, the system checks whether those orders fit within the available capacity of each work center. If not, the planner must adjust the plan - moving orders, adding overtime, or subcontracting. The adjusted plan is fed back into MRP, which recalculates material requirements based on the feasible schedule. |
The second feedback mechanism is execution feedback. As work orders are released and completed on the shop floor, the actual completion dates are recorded. If a work order finishes late, the ERP updates the status of dependent orders. A late completion might cause downstream shortages, triggering expedite messages. The closed loop ensures that MRP always reflects reality, not just the original plan. |
Closed-loop MRP is the precursor to modern ERP. It recognized that planning and execution are not separate activities but a continuous cycle. The plan informs execution. Execution data feeds back into the plan. The loop closes, and the factory learns and adapts. |
|
5.10 The Limits of MRP - What It Cannot Do |
MRP is brilliant at what it does, but it has limits. Understanding these limits is essential to using MRP well. |
MRP assumes infinite capacity. It will happily generate a plan that requires one hundred hours of work on a machine that only has eighty hours available. The plan is valid from a material perspective but impossible from a capacity perspective. This is why capacity planning - Rough-Cut Capacity Planning at the MPS level and more detailed capacity requirements planning at the work order level - is an essential complement to MRP. |
MRP assumes fixed lead times. In reality, lead times are not fixed. If a machine is overloaded, the lead time for parts processed on that machine increases. MRP does not capture this effect automatically. A planner must recognize the overload and adjust lead times or reschedule orders. |
MRP assumes perfect data. If the BOM is wrong, or inventory counts are inaccurate, or lead times are misstated, MRP's outputs will be wrong. Garbage in, garbage out. This is why BOM accuracy and inventory accuracy are such high priorities in factories that use MRP effectively. |
MRP assumes that orders can be split and moved arbitrarily. In reality, there are minimum order quantities, batch size constraints, and setup times that favor producing certain quantities together. MRP can be configured to respect these constraints, but doing so requires careful parameter settings. |
Despite these limits, MRP remains the most powerful material planning tool ever devised. No other approach comes close in its ability to coordinate thousands of components across complex product structures. |
|
5.11 MRP in the Age of Lean and Kanban |
In the 1980s and 1990s, the lean manufacturing movement, pioneered by Toyota, challenged some of MRP's assumptions. Lean advocates argued that MRP's detailed, centralized planning was unnecessary. Instead, they proposed kanban - a simple, visual pull system where downstream processes signal upstream processes when they need more parts. Kanban is elegant, low-cost, and very effective for repetitive, high-volume manufacturing. |
But MRP did not disappear. Instead, smart factories learned to use MRP and kanban together. MRP handles the long-term planning: forecasting, purchasing raw materials with long lead times, and managing capacity. Kanban handles the day-to-day execution: moving parts between work centers, controlling work-in-progress levels, and smoothing flow. |
In a hybrid system, MRP might create a weekly schedule for a machining cell, telling it what parts to produce and in what quantities. Within that week, the cell uses kanban to manage the flow of parts between machines. The ERP system supports both. It runs MRP to generate the high-level plan, and it manages kanban loops for the detailed execution. |
This hybrid approach is now common in mechanical manufacturing. MRP provides the brain - the strategic, long-range planning. Kanban provides the reflexes - the fast, local adjustments. Together, they create a factory that is both planned and responsive. |
|
5.12 Real-World Example: The Valve Manufacturer's MRP Transformation |
Consider a manufacturer of industrial valves. Each valve contains dozens of machined parts, purchased seals, springs, and fasteners. The company had a hundred product families and thousands of unique parts. Before MRP, they planned using a manual spreadsheet system. A team of five planners spent their entire week calculating requirements, placing purchase orders, and chasing shortages. Despite their efforts, shortages were common. The factory often stopped because a simple seal or spring was missing. |
The company implemented a modern ERP with a robust MRP engine. The first step was cleaning the data. They spent three months auditing BOMs, correcting inventory counts, and verifying lead times with suppliers. This was painful, but essential. Then they ran their first full MRP explosion. |
The results were shocking. The manual system had consistently over-ordered some parts and under-ordered others. They discovered a simple brass fitting that they had been ordering monthly in quantities of one thousand, when the actual need was only two hundred per month. The excess fittings filled four shelves. Meanwhile, a special alloy steel bar, critical for high-pressure valves, had been under-ordered for six months. The shortage had caused multiple production stoppages, but no one had traced the stoppages to the missing bar. |
With MRP running weekly, the planners were freed from manual calculation. Their role shifted from data entry to exception management. They spent their time acting on action messages - expediting late supplier orders, rescheduling work orders, and investigating discrepancies. Shortages dropped by eighty percent. Inventory levels fell by thirty percent, because they no longer ordered parts they did not need. On-time delivery rose from seventy percent to ninety-five percent. |
The transformation was not automatic. It required disciplined data management and a willingness to trust the system. But once the factory learned to feed MRP good data and act on its outputs, the classic brain performed exactly as its inventors had envisioned. |
|
5.13 Summary: The Brain That Never Forgets |
Material Requirements Planning is called the classic brain because it is the intellectual engine that makes modern mechanical manufacturing possible. Before MRP, factories were limited to simple products and long lead times. The complexity of a modern gearbox, pump, or engine would have been impossible to plan manually. There were simply too many parts, too many relationships, and too many timing constraints. |
MRP changed everything. It replaced hunches with calculations, guesses with logic, and chaos with coordination. It does not require genius. It requires only accurate data and a computer that can perform simple arithmetic billions of times without error. MRP's logic is not hidden or proprietary. It is open, transparent, and teachable. Any planner can learn how MRP thinks, and any factory can implement it. |
But MRP is not a substitute for human judgment. It is a tool that amplifies human judgment. The planner still decides how much safety stock to hold, how to respond to an expedite message, and whether to trust a supplier who has been late three times in a row. MRP provides the data. The planner provides the wisdom. |
In the next chapter, we will explore how the ERP adds capacity planning to the material planning we have covered here. Material without capacity is a plan that cannot be executed. Capacity without material is a factory that cannot build. Together, they form the complete planning picture. But the heart of that picture - the classic brain - will always be MRP. |
|
Key takeaways from Chapter 5: |
1. MRP replaced guesswork with a logical, algorithmic approach to material planning, transforming manufacturing in the 1960s and 1970s. |
2. MRP uses three inputs: the Master Production Schedule, the Bill of Materials, and inventory status. |
3. BOM explosion multiplies requirements through product structures, respecting dependent demand relationships. |
4. Time-phased netting subtracts on-hand inventory and scheduled receipts, but only if the timing aligns with requirements. |
5. Low-level coding solves the problem of components that appear in multiple products at different BOM levels. |
6. Safety stock provides a buffer against uncertainty but increases inventory costs - the ERP calculates suggested levels, but management chooses the trade-off. |
7. Action messages (release, expedite, de-expedite, cancel, reschedule) focus the planner's attention on what matters most. |
8. Closed-loop MRP adds capacity feedback and execution feedback, making the plan realistic and current. |
9. MRP has limits: it assumes infinite capacity, fixed lead times, perfect data, and flexible batch sizes - smart planners work around these limits. |
10. MRP and lean kanban are complementary: MRP for long-term planning, kanban for day-to-day flow. |
11. Real-world success requires clean data and disciplined execution - MRP is powerful but not forgiving of garbage input. |
cs@easiersoft.com