Is One-Piece Flow Always Better?

Short answer: If the question is flow versus large batch production or stagnation, then yes — flow is always better. But the question as commonly asked conflates several things. Flow (nagareka, 工程の流れ化) is one of the basic principles of JIT, not the whole system. It works alongside takt time, pull, and leveling. And the more specific question — one-piece flow versus small-lot production — is where the real distinction lies.

Is flow always better than batch?

Yes. Flow versus large batch production or stagnation is generally not a close call. Flow shortens lead time, exposes problems immediately, reduces inventory, and connects processes so that downstream problems surface upstream. This is foundational in TPS. In discrete parts and automotive types of industry, it is not even debatable.

But flow alone is not sufficient. This is a common pattern in lean questions — some questions need reframing to give the fuller answer.

Toyota’s TPS glossary (用語集) places continuous flow processing (工程の流れ化, kōtei no nagareka) as one of three basic principles of Just-In-Time, with leveling (heijunka, 平準化) as the premise. The three principles are:

Downstream pull (後工程引き取り, atokōtei hikitori) — the next process withdraws what it needs from the previous process
Continuous flow processing (工程の流れ化) — eliminating stagnation within and between processes
Takt — determining the pace of production from the required quantity

Flow without takt means you are flowing at whatever speed suits the process, not the customer. Flow without pull means you are pushing material into downstream processes regardless of need. Flow without leveling means demand spikes propagate through the system and destroy the flow you built. All four work as a system. Asking “is flow better?” without the other three is like asking if an engine is better than no engine — yes, but it helps to have a transmission, chassis, wheels, body frame, etc.

What is the difference between one-piece flow and small-lot production?

Within the flow principle, there is a further distinction. The TPS glossary defines one-piece-at-a-time production (1個流し, ikko nagashi) as “processing or assembling one piece at a time in process order, and flowing one piece at a time to the next process.” The ultimate goal is making things one by one with 100% quality to customer specification on demand.

That is the ideal state. The current-state reality across many processes — including inside Toyota — is small-lot production. The two are not in conflict. Small-lot production is the expression of the flow principle within the constraints that exist today.

Some processes are constrained by physics. Stamping lots at Toyota run in the hundreds of parts because die changeovers, even after decades of setup reduction, require a minimum lot size. The heat treat lot size at Kamigo engine plant was 40 parts when I worked there — heating one part at a time in a furnace would be a terrible waste of energy. Casting and other processes often have similar thermal and physical constraints. These are governed by what Toyota calls signal kanban (信号かんばん, shingō kanban) in TPS — a triangle-shaped kanban that triggers the next production lot at a reorder point. These are not failures of TPS maturity. They are often the laws of physics applied to the current state of technology.

The other dimension is evolution. Lot sizes at Toyota have gotten smaller over decades as setup reduction, equipment design, and process technology improved. What was a batch of thousands in the 1960s became hundreds. Set up times plunged from an average of 4 hours in the 1950’s to 15 minutes in 1962 and under 3 minutes in the 1970’s. In the future, what was hundreds may become tens as technology continues to advance. The direction is towards one-piece flow. The constraint is what physics and economics allow today.

This is the distinction that gets lost. The question “is one-piece flow always better?” sets up a false choice between flow and batch. Less waste is better. Less stagnation and better flow is improvement. One-piece flow is the ideal expression of flow. Small-lot production is where many processes actually operate — including inside Toyota — and the work is to keep pushing lot sizes smaller over time as conditions allow. The preconditions for getting closer to one-piece flow — equipment reliability, quality at source through jidoka and poka-yoke, setup reduction, standardized work, and leveling — are where the real improvement work lies.

Is flow always better than batch?

What is the difference between one-piece flow and small-lot production?

Related