Lean-MRP Hybrid: How to Combine Pull Scheduling with MRP for Long-Lead Materials

Three years into a lean transformation, a mid-size automotive parts manufacturer had fixed their shop floor. Kanban cards ran smoothly between cells. Takt time was calibrated. Changeover times were down 60%. Throughput was up. Then they ran out of a key machined blank from a Korean supplier. Lead time: 14 weeks. Their kanban system had no mechanism to signal that far ahead.

The shop floor was lean. The supply chain was not. And lean had given them no tool for the part of the problem that kanban cannot reach.

This tension—between lean's pull philosophy and MRP's push logic—is real and poorly understood. Most lean literature presents them as incompatible. In practice, most successful lean manufacturers use both, in a deliberate hybrid architecture that assigns each tool to the part of the supply chain where it actually fits.

After 35 years supporting manufacturers ranging from job shops to Tier 1 aerospace suppliers, User Solutions has helped production teams navigate exactly this transition. Here is the architecture that works.

For full MRP context, see our complete MRP guide and our post on MRP vs. MRP II vs. ERP.

Why Lean and MRP Appear to Conflict

The lean critique of MRP is specific: MRP is a push system. It generates a schedule based on a plan, then pushes work onto the shop floor whether downstream operations are ready or not. This creates work-in-process accumulation at bottlenecks, inflates lead times, obscures problems, and makes throughput dependent on schedule accuracy rather than actual demand.

Kanban and pull scheduling solve this by letting downstream operations signal their readiness. Work moves only when the next station calls for it. WIP is capped. Problems surface immediately. Lead times compress because queues disappear.

The lean argument is correct—for the shop floor. The problem is that it does not extend cleanly to supply planning for purchased materials.

Kanban works for purchased components when:

• Supplier lead time is short relative to your production cycle (days, not weeks)
• Demand for the component is stable and predictable
• You have physical space for a supermarket (buffer stock) near point of use
• You can make the signal-to-delivery loop fast enough to avoid stockouts

When supplier lead times are 6, 10, or 14 weeks, and demand is variable or custom-configured, kanban alone cannot bridge the gap. The signal from the shop floor kanban triggers a reorder today, but the material won't arrive for 14 weeks. If your production plan changes in week 8, you either eat the inventory or you expedite.

The Actual Role of MRP in a Lean Environment

MRP should not govern your shop floor in a lean environment. But it should govern your purchased material supply chain—specifically for components where:

1. Supplier lead time > customer's acceptable lead time: You must plan ahead because you cannot respond to customer demand in real time. Only MRP (or equivalent forward planning) can generate the right purchase orders at the right time.

2. Demand is non-repetitive or highly variable: Kanban sizing assumes relatively stable demand. For custom or engineered components where order quantities vary significantly by job, MRP calculates the right quantity for each period.

3. Components are product-specific: A machined casting unique to one product family cannot be replenished by a generic kanban card—it needs demand-driven planning tied to the specific orders driving it.

For everything else—high-volume, stable-demand, common components with short supplier lead times—kanban replenishment is simpler, faster, and more robust than MRP.

The practical division in most lean-MRP hybrid plants:

Decoupling Points: The Design Decision That Makes It Work

The lean-MRP hybrid works because of decoupling points—inventory buffers that separate the MRP-governed supply side from the kanban-governed production side.

At a decoupling point, you hold a physical inventory buffer (often called a supermarket). The shop floor pulls from this supermarket via kanban signals. MRP replenishes the supermarket based on a forward-looking production plan, ahead of when the kanban will actually need the material.

The buffer absorbs the mismatch between:

• MRP's planning horizon (weeks to months ahead, based on forecasts and customer orders)
• Kanban's signal horizon (days, based on actual consumption)

Without a decoupling point, every kanban signal must propagate all the way back through MRP lead times to a supplier. With a decoupling point, the kanban signal propagates only to the supermarket. MRP handles the supplier relationship independently, using the production plan to project future supermarket depletion.

Choosing Where to Place Decoupling Points

The primary criterion: place decoupling points where the material lead time structure changes significantly.

At purchased component supermarkets: The most common decoupling point. Components with long supplier lead times are held in supermarket buffers near point of use. MRP replenishes the supermarket weekly or biweekly based on the production plan. The shop floor kanban pulls from the supermarket.

At sub-assembly level (for complex products): For products with multi-level BOMs, a second decoupling point at the sub-assembly level can buffer variability in both the purchased component supply and the sub-assembly build process, giving final assembly a stable, reliable feed.

At finished goods (for make-to-stock): For standard products replenished to stock, the finished goods inventory itself is the decoupling point between customer demand and production.

The number of decoupling points should be the minimum needed to absorb lead time and variability mismatches. Every decoupling point is an inventory investment. Over-decoupling creates waste; under-decoupling creates shortages.

Sizing the Buffer at Each Decoupling Point

Buffer sizing is where the lean-MRP hybrid gets technical. The buffer must be large enough to cover demand during the replenishment lead time—with a margin for variability—but small enough to minimize working capital.

A practical sizing approach:

Step 1: Calculate average daily usage (ADU) of the component over the past 90 days. Use actual consumption, not forecast.

Step 2: Identify the replenishment lead time — the calendar days from MRP generating a purchase order to the component arriving in the supermarket (includes PO processing time + supplier lead time + receiving/inspection).

Step 3: Set minimum buffer = ADU × replenishment lead time. This is the quantity consumed during the replenishment cycle if demand holds at average.

Step 4: Add variability factor: Multiply the minimum buffer by 1.3–1.7 depending on demand volatility and supplier reliability. Higher variability = larger buffer.

Step 5: Set maximum buffer = minimum buffer × 2 (roughly). Above the maximum, you are holding excess. Below the minimum, you are at risk of stockout.

Example:

• Part: machined aluminum bracket, used in 3 product families
• ADU: 45 units/day
• Replenishment lead time: 6 weeks = 42 calendar days
• Minimum buffer: 45 × 42 = 1,890 units
• Variability factor: 1.4 (moderate demand variability, good supplier)
• Target buffer: 1,890 × 1.4 = 2,646 units (round to 2,700)
• Maximum: 2,700 × 2 = 5,400 units (trigger to slow purchasing if above this)

MRP monitors actual buffer inventory and generates replenishment orders when the buffer drops toward the minimum. The shop floor kanban signals never interact with MRP directly—they interact only with the physical supermarket.

DDMRP: The Formalized Version of This Approach

Demand-Driven MRP (DDMRP), developed by the Demand Driven Institute and popularized by authors Chad Smith and Carol Ptak, formalizes the lean-MRP hybrid into a structured methodology.

DDMRP introduces:

• Five-step positioning methodology: rules for selecting decoupling points based on lead time, variability, and strategic importance
• Buffer sizing formula: combines ADU, lead time, and variability into three buffer zones (red = danger, yellow = working, green = healthy)
• Demand-driven alerts: MRP generates replenishment signals based on actual demand consumption against the buffer, not against a forecast
• Relative priority: planning priorities are visible as buffer color status rather than date-based exception messages

DDMRP is particularly well-suited to:

• Environments with high mix and variable demand
• Multi-level BOMs with long supplier lead times at lower levels
• Companies that have done lean on the shop floor and now need to extend it to purchasing

It is not a replacement for MRP logic—it is MRP logic redesigned around decoupling points and buffer management rather than dependent demand explosion from a fixed plan.

The MRP-to-Kanban Interface: What It Looks Like Operationally

In a functioning lean-MRP hybrid, here is what the daily workflow looks like for your planning and purchasing team:

Weekly MRP run:

1. MRP ingests the current production plan (customer orders + forecasts for the horizon beyond your frozen zone)
2. For long-lead purchased components, MRP explodes demand through the BOM and compares against current supermarket inventory
3. Where projected inventory drops below buffer minimum within the supplier lead time, MRP generates a planned purchase order
4. Purchasing reviews planned POs, converts to actual POs, and sends to suppliers
5. No MRP-generated work orders are released to the shop floor—the shop floor is governed by kanban

Daily shop floor:

1. Production pulls from supermarket using kanban cards (physical or electronic)
2. Supermarket inventory decreases as kanban cards are consumed
3. Kanban replenishment signal (card returns to designated replenishment point) is logged in the system—this updates the supermarket inventory record in real time
4. MRP sees the current supermarket level on the next run (or in real time if the system supports it) and adjusts planned POs accordingly

When a customer order changes:

• If the change is in the open planning zone: MRP adjusts planned POs on the next run automatically
• If the change is in the slushy zone: MRP flags the impact for planner review
• If the change is in the frozen zone: the kanban system continues to run unaffected; the planner evaluates whether any already-ordered long-lead material needs to be redirected or cancelled

Common Failure Modes in Lean-MRP Hybrid Implementations

Failure 1: Using MRP to push work onto the shop floor

The most frequent hybrid failure: implementing MRP for supply planning, then also using MRP-generated work orders to drive shop floor sequencing. This destroys the pull system. Work orders from MRP will not align with the pull signals from downstream operations. WIP builds up. The kanban discipline breaks down within weeks.

Rule: MRP generates purchase orders only. Work orders are triggered by pull signals (kanban, CONWIP, or equivalent). Keep these systems strictly separate.

Failure 2: Not building physical supermarkets

The decoupling point only works if there is a physical location where the buffer is held and managed. Companies that implement the theory without creating the physical supermarket location (dedicated space, clearly labeled, FIFO-managed) lose visibility into actual buffer levels and the whole system degrades.

Failure 3: Setting buffer sizes from gut feel

Buffers set too small cause stockouts that disrupt lean flow. Buffers set too large consume working capital and hide problems. Use actual consumption data and actual supplier lead time data (not quoted lead times—see our post on MRP data accuracy) to size buffers properly.

Failure 4: Running MRP daily on a lean shop floor

Daily MRP re-runs in a hybrid environment generate noise. Long-lead purchased materials should be planned weekly or biweekly—daily replanning rarely improves output quality and often generates conflicting signals that confuse purchasing. Reserve daily replanning for genuinely high-velocity, short-lead environments.

When to Add More MRP, When to Add More Pull

As your hybrid matures, you will periodically face the question of whether to expand MRP coverage or expand the pull system. The answer depends on the lead time and demand variability profile of the specific components in question.

Expand MRP when: You are adding new product lines with longer-lead specialty components. Supplier lead times in your industry are increasing (common in 2024–2025 conditions). You are moving into engineer-to-order products with unique material requirements.

Expand pull when: A component that was once custom is becoming standardized and common. A supplier has dramatically reduced their lead time. A component has high enough volume to support a dedicated kanban loop. You are simplifying your product line and moving toward fewer configurations.

A lean-MRP hybrid is not a static architecture. It evolves as your product mix, suppliers, and market conditions change. The best planning teams review the assignment of components to MRP vs. pull annually—moving components between systems as their characteristics change.

---

---

Trying to make lean and MRP work together? Contact User Solutions to see how RMDB supports hybrid pull-push scheduling environments—with MRP for long-lead purchasing and pull scheduling for shop floor execution. Trusted by GE, Cummins, BAE Systems, and hundreds of manufacturers for 35+ years.