411. Decoupling Stock

Inventory held between two production stages (or between a supplier and a buyer) so each stage can operate independently — without being immediately blocked by problems in the upstream stage.

Conceptually similar to safety stock, but for internal stages rather than external customer demand.

411.0.1. The problem decoupling solves

Consider a two-stage line: stage A feeds stage B. If they operate strictly tied (no buffer between):

Total throughput is bounded by the minimum of the two stages’ instantaneous availability — not the average. This compounds dramatically with more stages.

Decoupling stock = a buffer between A and B. A pile up parts; B pulls from the pile.

Each stage can handle its own short-term variation independently.

411.0.2. When decoupling matters most

The cost of not decoupling depends on:

Decoupling stock is most valuable when:

It’s least valuable in a perfectly flow-balanced, ultra-reliable, automated environment — exactly the conditions Toyota engineered for, which is why TPS deliberately reduces decoupling stock to expose problems.

411.0.3. Toyota’s view: stock hides problems

Lean / TPS treats decoupling stock as waste. A buffer between A and B disguises A’s reliability problems — the line keeps running, but you never feel pressure to fix the root cause.

Toyota’s approach:

The famous “lower the water to expose the rocks” metaphor: stock = water level, problems = rocks. Lean reduces stock to surface problems and force solutions.

411.0.4. Sizing decoupling stock

Various models:

Closed forms exist for some special cases (e.g., M/M/1 stages); more often simulated.

411.0.5. How it composes

Component Magnitude Where it lives
Cycle stock At each stocking location
Safety stock At each stocking location
Pipeline stock In transit between stages
Anticipation stock planned Centralized for known events
Decoupling stock varies (insurance / SS-like) Between production stages — physically a queue or buffer area
Example: Two-stage manufacturing line

Given:

  • Stage A: machining. Cycle time 60 sec/unit. Average downtime 5% (3 min/hour).
  • Stage B: assembly. Cycle time 60 sec/unit. Reliable (negligible downtime).
  • Both stages run 8 hours/day.

Step 1 — without decoupling

Tied stages: A’s downtime immediately idles B. Effective throughput = × nominal capacity (A’s availability dominates because B has no buffer).

At 60 sec/unit: 480 units/day at full capacity → 456 units/day with 5% A-downtime.

Step 2 — small decoupling buffer

Add a 30-unit buffer between A and B.

Now A’s downtime affects B only if A is down longer than the buffer can sustain B (30 units × 60 sec = 30 minutes).

In a typical day, A’s downtime is split into many short events (e.g., 12 events × 15 sec each rather than one 3-minute outage). A 30-unit buffer easily absorbs all of them. Effective throughput approaches 480/day.

Cost: holding 30 units of WIP at, say, $20/unit ⇒ $600 capital + small holding cost. Throughput gain: 24 units/day ⇒ over a year, extra units valued at margin.

Step 3 — Toyota approach: shrink the buffer

Toyota would target zero buffer — and then attack the 5% A-downtime root cause until it disappears (preventive maintenance, redesign, jidoka). Once A is reliable enough, the buffer can come out.

The general decoupling-stock decision

At each stage boundary:

  1. Estimate the cost of stage starvation (lost throughput × margin).
  2. Estimate the cost of holding the buffer.
  3. Set buffer size to balance.
  4. (If pursuing lean): reduce variability at upstream stages until the buffer is no longer needed, then remove it.

Decoupling stock is the only inventory category Toyota systematically eliminates rather than optimizes.