Demand spikes destroy process capability. When a plant lands a major order, standard practice is to authorize overtime, pull maintenance teams from preventive work, and push throughput. Within days, nonconformance reports flood the quality department, WIP clogs the assembly lines, and the customer whose order triggered the surge calls to ask why their first shipment is already late. I have audited plants where this exact cycle is accepted as standard operating procedure.
Organisations confuse this reactivity with productivity. Demand surges, production strains to meet it, quality degrades, and personnel burn out. When demand inevitably drops, the same factories send operators home early while management wrings its hands over idle capacity. Toyota identified this systemic failure in the mid-20th century and implemented Heijunka, or production leveling, as a foundational pillar of the Toyota Production System to combat it.
Toyota deliberately produced below peak capacity during high-demand periods and maintained output during slow periods. They recognised that unevenness in production is not merely a logistics problem. It is an expensive, unmeasured quality problem that generates variation and defects across the entire value stream.
The Mechanics of Volume and Mix Leveling
Heijunka requires manufacturing goods at a steady, consistent rate rather than processing large batches dictated by immediate customer orders. The objective is to smooth fluctuations in both total volume and product mix, ensuring that daily operations remain predictable. This stability is what allows quality systems to function correctly.
Volume leveling produces a fixed total quantity of units per shift. Instead of building 1,000 units on Monday and 300 on Tuesday, a facility builds 650 on both days. Mix leveling distributes product variants evenly across the schedule, interleaving variants (A, B, A, B) rather than running massive batches of a single product. This prevents downstream operations from starving and keeps the workforce alert to specification changes.
Together, these two dimensions create a production rhythm that directly supports quality control. Predictable cadence gives operators the mental bandwidth to execute standardised work accurately. It eliminates the cognitive overload that occurs when a facility constantly ramps production up and down, directly reducing the human error component of defect generation.
How Uneven Production Destroys Process Capability
The relationship between scheduling and defect rates is profound, yet most quality professionals focus strictly on inspection effectiveness and Cpk. When production surges, every resource is stretched beyond its validated capacity. The operator running two machines is forced to run four. The inspector who requires five minutes per part is given two. Quality degrades because the system exceeds its physical limit to maintain standards.
Uneven production also triggers a bullwhip effect in the supply chain. Erratic demand signals force suppliers to ramp up and shut down abruptly, introducing variation in raw materials. Different lots, varying lead times, and fluctuating storage conditions arrive at your dock as unplanned sources of variation. This upstream chaos directly undermines your PFMEA assumptions and destabilises your process.
Large-batch scheduling forces massive WIP inventory that sits between operations, aging and degrading. When components sit too long, they risk contamination or get mixed with newer batches. Processing material of uncertain vintage through a validated process inevitably leads to nonconformances. Furthermore, the fatigue associated with extended overtime shifts consistently drives up accident and defect rates.
The Mathematics of Stable Output
Consider two facilities producing the same 10,000 units per week. Plant A produces based on immediate demand: 3,000 units on Monday, tapering to 1,000 on Thursday, and 2,000 on Friday. The peak-to-trough ratio is 3:1. On peak days, machinery maxes out, maintenance is deferred, and inspection is rushed. On slow days, capacity sits idle. This swing generates massive variation.
Plant B produces 2,000 units every day. The total weekly output is identical, but the peak-to-trough ratio is 1:1. Staffing remains stable, preventive maintenance occurs as scheduled, and inspections are properly paced. Suppliers deliver consistently, ensuring raw material stability. Plant B will reliably achieve higher process capability and lower internal scrap rates.
The Plant A vs. Plant B Scheduling Profile
Most manufacturers operate closer to Plant A, believing they are being responsive to customer demand. They are simply being reactive. Reactivity forces a system to operate outside its validated parameters, making consistent quality impossible to achieve regardless of the inspection regime in place.
Changeover Frequency and SMED Integration
In traditional large-batch scheduling, changeovers are high-stakes events. A setup error can produce hundreds of defective parts before the next scheduled first-article inspection. Organisations invest heavily in setup verification to prevent these massive scrap events. Because changeovers are risky, schedulers actively avoid them, reinforcing the batch-and-queue mentality.
Heijunka forces smaller, more frequent batches, which means more changeovers. However, the risk per transition drops dramatically because the batch at risk is smaller. Furthermore, frequent changeovers force operators to master the setup procedure. Quality improves because the changeover becomes a routine, standardised operation rather than an exceptional, high-pressure event.
This is why Heijunka and Single-Minute Exchange of Die (SMED) are inseparable. Leveling creates the operational need for fast changeovers, and SMED provides the engineering method to achieve them. Together, they transform a quality-threatening disruption into a predictable, low-risk operational rhythm that maintains flow without sacrificing first-time yield.
Reactivity forces a system to operate outside its validated parameters, making consistent quality impossible to achieve.
Implementing Heijunka as a Quality Tool
Implementing production leveling is not a scheduling exercise; it is a systemic quality initiative. It begins with establishing your takt time—the rate at which you must produce to meet average customer demand. Takt time dictates the pace of the entire value stream and forms the baseline for all standardised work. Without a firm takt, leveling is impossible.
Next, apply SMED methodologies to reduce changeover times. If equipment transitions take four hours, you cannot level your product mix. Engineering must design tooling and setup procedures that take minutes, not hours. Once changeovers are under ten minutes, scheduling mixed-model production becomes practical and sustainable without halting the line.
Finally, establish a small, deliberate finished-goods buffer. This buffer absorbs the variation between your steady production rate and the customer's variable demand pattern. It replaces the chaotic, massive WIP inventories of batch production with a managed stock of finished goods. The facility produces at a consistent rate, shipping to the customer from the buffer, entirely decoupling your internal quality processes from external demand spikes.
Establishing a Heijunka Production Rhythm
- 01Calculate Takt TimeDetermine the steady production rate required to meet average customer demand over a fixed period.
- 02Reduce ChangeoversApply SMED principles to ensure product mix can be changed without disrupting the production flow.
- 03Build Finished BufferEstablish a small, managed inventory to absorb customer demand variation and shield the production floor.
- 04Level the MixDistribute product variants evenly across the schedule to prevent downstream starvation and maintain operator focus.
The most persistent objection to this approach is the fear of increased inventory. Leveling will slightly increase finished-goods inventory, but it dramatically reduces WIP and raw-material inventory. More importantly, it eliminates the hidden inventory of defects and scrap generated by overloaded processes. The net financial and quality impact is overwhelmingly positive.
Measuring the Impact on Defect Rates
To prove the business case, map your internal defect rate against your production volume over the past six months. If your scrap and rework percentages rise and fall in direct correlation with your production volume, you have a scheduling problem masquerading as a quality problem. This correlation is visible in the 8D reports generated during peak periods.
Organisations that plot this data consistently find that their lowest defect rates occur during their most stable production weeks—not the slowest or fastest weeks, but the most level. Process capability improves because the system operates within its validated parameters. Preventive maintenance is executed on time, material flow is consistent, and operators maintain a sustainable pace.
Defects decrease under Heijunka because fewer of them are created in the first place. The system simply stops generating the overload, fatigue, and material variation that cause nonconformances. Production leveling is arguably the most underrated quality tool available because it addresses the root cause of variation directly at the scheduling level, long before the product ever reaches an inspection station.