Heijunka vs Mixed-Model Production: Two Lean Words for the Same Goal With Different Mechanics

Key takeaways

• Heijunka = leveling production volume and mix over a time window. The goal is smooth, predictable flow.
• Mixed-model production = running multiple SKUs through the same line, often in interleaved sequences.
• Heijunka often implements via mixed-model production, but they are not identical concepts.
• Both fight the same enemy: batch-driven demand volatility that produces over-and-under-production cycles.
• Plants doing one without the other often miss the bigger benefit available from doing both.

Short answer: Heijunka is the lean practice of leveling production volume and mix over a time window so that flow becomes smooth and predictable. Mixed-model production is the operational technique of running multiple SKUs on the same line in interleaved sequence. Heijunka is often implemented via mixed-model production. They overlap heavily but are not identical — heijunka is the goal, mixed-model is one of the tools.

What heijunka is

Heijunka (平準化) means "leveling" in Japanese. The practice levels both volume and mix:

• Volume leveling. Production rate is steady over time. No spike days, no slack days.
• Mix leveling. SKU mix is spread evenly over the time window. Not all of SKU A on Monday and all of SKU B on Tuesday.

The goal is smooth, predictable flow that does not amplify demand variation back through the supply chain.

What mixed-model production is

Mixed-model production runs multiple SKUs through the same line in interleaved sequence. Rather than running 200 of SKU A then 200 of SKU B, the line runs A-B-A-B-A-A-B-A-B-B etc.

This requires:

• Short changeover (SMED) so SKU transitions do not destroy throughput.
• Sequence-aware material delivery (right parts at right time).
• Operator flexibility on multiple SKUs.
• OEE measurement that tolerates SKU mix.

How they relate

Heijunka is the goal: level the production. Mixed-model production is one way to achieve it. There are others:

• Fixed sequence. Same SKU pattern repeats every day. Levels mix but not within a day.
• Frequency leveling. SKUs that need a lot run more often; SKUs that need less run less often. No interleaving required.
• Pure mixed-model. Every interleaved unit reflects demand mix.

For most plants pure mixed-model is the strongest version of heijunka, but it requires the strongest enablers (very short changeover, flexible operators, sequence-aware logistics).

Why heijunka matters

Unleveled production creates:

• Bullwhip effect. Demand variation amplifies upstream through the supply chain.
• Inventory swings. Build up during high days; deplete during low days.
• Operator stress. Cycle ramps and downs strain workforce.
• Quality variability. Setup at high cadence vs steady-state at low cadence creates inconsistent process conditions.

Heijunka eliminates these.

The trade-off heijunka requires

Leveling requires:

• Demand smoothing or predictable demand windows.
• Short changeover capability.
• Material readiness for multiple SKUs.
• Operator flexibility.

Plants without these enablers cannot run pure heijunka. They settle for partial leveling — typically weekly fixed sequence or frequency leveling.

How mixed-model production affects OEE

Mixed-model production usually:

• Lowers single-SKU throughput peaks (more changeovers).
• Reduces inventory swings.
• Reduces overall lead time.
• Stabilizes Quality (consistent conditions).

OEE on a mixed-model line typically has more changeover loss in Availability but better Performance stability and Quality. Net OEE depends on changeover speed (which is why SMED is the prerequisite).

Common mistakes

1. Calling weekly schedule "heijunka." Weekly schedules level over weeks but not within them. Real heijunka is daily or sub-daily leveling.

2. Mixed-model without SMED. Constant changeovers eat all the throughput. Mixed-model requires single-digit-minute changeover.

3. Heijunka with no demand smoothing. Wild demand cannot be perfectly leveled. Heijunka levels against forecasted average; high variance breaks it.

4. Treating heijunka and mixed-model as substitutes. One is the goal, the other is one technique to achieve it.

When fixed-sequence or frequency leveling is better

• Long changeover times that SMED cannot reduce enough.
• SKU-specific tooling that makes interleaving expensive.
• Small product mix where pure interleaving is overkill.

How a modern OEE platform supports heijunka

A modern OEE platform tracks per-SKU OEE on mixed-model lines, exposes changeover time, surfaces the cost of mix changes, and helps the scheduler find leveled sequences that maintain throughput.

Fabrico's OEE module supports per-SKU OEE on mixed-model lines, surfaces changeover impact on overall OEE, and reports the trade-off between leveled flow and throughput losses.

See how Fabrico captures this automatically — explore OEE for manufacturing or book a demo.

Related reading

Frequently asked questions

Is heijunka the same as mixed-model production?

No. Heijunka is the goal of leveled production; mixed-model is one way to achieve it.

Do I need SMED before mixed-model production?

Usually yes. Frequent changeovers without SMED destroy throughput.

Can I do heijunka without leveling mix?

Volume-only leveling is partial heijunka. Mix leveling is the harder and more valuable part.

Does heijunka work with high demand variability?

Partially. The base demand can be leveled; spikes still need buffering.

Is heijunka compatible with kanban?

Yes — they often pair. Heijunka levels the production schedule; kanban signals the actual replenishment.