Most automotive and aerospace plants do not have a capacity problem; they have a rhythm problem. Assembly lines sprint at maximum output early in the week, generating overtime and quality shortcuts, before grinding to a halt on Thursday. This oscillation between feasting and fasting destroys process stability and inflates internal lead times.
This volatile scheduling is the exact opposite of Heijunka, the production leveling discipline developed within the Toyota Production System. Heijunka distributes both production volume and product mix evenly across a defined time horizon. It is the prerequisite for Just-In-Time (JIT) delivery, because you cannot run a pull system when your schedule fluctuates by three hundred percent overnight.
Implementing production leveling requires shifting focus from local workstation efficiency to total system flow. I have audited plants where leaders maximised OEE on individual machines, only to find massive hidden WIP inventory and missed customer deliveries. Heijunka sacrifices local efficiency to achieve system-level stability.
The Mechanics of Volume and Mix Leveling
Leveling operates across two dimensions: volume and mix. Volume leveling replaces boom-and-bust scheduling with a fixed daily quantity. Instead of building 4,800 units in three days and idling, a facility produces 1,370 units every day. The total weekly output remains identical, but the operational pace becomes sustainable.
Mix leveling is the deliberate interleaving of product variants to match actual demand. A traditional planner running products A, B, and C will run long batches of each to minimise changeovers. This approach maximises local machine efficiency but devastates downstream flow and artificially extends lead times for lower-volume variants.
A leveled schedule reverses this logic. If demand is 60% product A, 30% product B, and 10% product C, the production sequence becomes A-A-B-A-A-B-C, repeating continuously. The math is simple, but the operational discipline required to execute it is immense.
Batch Scheduling vs. Heijunka Leveling
Batch-and-Queue Approach
- Maximises workstation OEE and utilisation
- Minimises direct changeover frequency
- Creates massive WIP inventory between processes
- Inflates overall production lead times
Heijunka Leveled Approach
- Prioritises total system flow over local speed
- Increases planned changeover frequency
- Drops WIP and finished goods inventory sharply
- Shrinks lead times and stabilises output
The Heijunka Box and Visual Management
The operational core of this discipline is the Heijunka Box, a scheduling tool that makes the production rhythm visually obvious. The box uses rows for each product family and columns for time increments, typically shifts or hours. Kanban cards representing specific production quantities are distributed evenly across the grid.
This visual management tool exposes deviations instantly. If cards are clustered heavily on Monday and absent on Wednesday, the schedule is not leveled. Modern Manufacturing Execution Systems (MES) can replicate this board digitally, but the core function remains identical: make the plan visible and make deviations unmistakable.
Visual transparency forces accountability. When operators, supervisors, and managers share the exact same view of the production intent, the daily friction between planning and execution drops. The board becomes the single source of truth, overriding ad-hoc requests that would disrupt the flow.
The Direct Impact on Process Quality
Production leveling is a core quality engineering tool, not just a lean scheduling technique. The mechanism connecting Heijunka to defect reduction is straightforward: smaller batch sizes provide faster feedback. When you build 4,800 units before changing over, a systematic process failure can run undetected for days. Interleaved batches catch the same defect after 200 units.
Stable, predictable schedules directly reduce human error. When operators are not fighting to hit unrealistic targets during peak days, they follow standard work. They do not bypass steps, and they do not force non-conforming parts through the line. This reduction in time pressure is essential for maintaining stringent aerospace and automotive quality standards.
Furthermore, frequent changeovers force frequent process verification. Every setup requires a first-piece inspection and parameter confirmation. In a batch system, you verify once and assume stability across thousands of units. In a leveled system, verification happens continuously, creating a dense network of quality checkpoints across the shift.
Finally, Statistical Process Control (SPC) mathematically assumes a stable process. A workstation running at maximum speed on Monday and sitting idle on Thursday generates wildly unstable variation. Heijunka enforces a consistent pace, ensuring the control limits calculated for your SPC charts are actually valid for the data being recorded.
Implementation Prerequisites and SMED
You cannot implement Heijunka in a chaotic environment. The first prerequisite is a stabilised demand signal. Analyse actual customer pull over a meaningful horizon, typically four to twelve weeks. If customer demand swings unpredictably by massive margins, you have a Sales and Operations Planning (S&OP) failure that must be addressed before attempting to level production.
The second prerequisite is aggressive changeover reduction. Interleaving products is economic suicide if changing a die takes four hours. You must apply Single-Minute Exchange of Die (SMED) methodologies to internal and external setup tasks. The target for every changeover must be under ten minutes to make the interleaved schedule viable.
Sequence for Implementing Production Leveling
- 01Stabilise DemandAnalyse actual customer pull over 4-12 weeks to find the true average takt time.
- 02Establish Basic FlowEliminate batch-and-queue steps; build defined pull systems using kanban.
- 03Apply SMEDDrive changeover times below ten minutes to enable frequent product interleaving.
- 04Execute the PatternRun the repeating sequence (e.g., A-A-B-A-A-B-C) and monitor adherence.
- 05Track DeviationsUse the Heijunka board to identify capacity bottlenecks and adjust ratios monthly.
Once prerequisites are met, define the initial production pattern. Start with two products in a fixed ratio. At WITTE Automotive, we found that running a simple, rigid pattern for four weeks allowed operators to adapt to the new rhythm before we introduced higher product variety. Complexity must be added gradually.
You cannot run Just-In-Time delivery if your production schedule looks like a roller coaster. JIT requires structural predictability.
Overcoming Local Optimisation Metrics
The primary barrier to Heijunka is not technical; it is behavioural. Leveling increases changeover frequency, which directly reduces per-shift efficiency at individual machines. Traditional cost accounting systems will flag this as a loss. Plant managers evaluated strictly on workstation OEE will abandon the effort before system-level benefits materialise.
This conflict requires executive leadership to redefine how performance is measured. You must shift the evaluation criteria from local machine utilisation to total order-to-delivery lead time. If leadership lacks the patience to absorb a temporary dip in local efficiency metrics, the culture will default back to massive batching.
This cultural shift demands a commitment to stability over adrenaline. Firefighting makes heroes out of shift leaders, but it destroys quality and inflates costs. An organisation built around solving daily crises will resist the steady, deliberate rhythm of a leveled schedule. Leadership must enforce the standard until the new rhythm becomes routine.
Measuring the Impact of Leveling
To verify that production leveling is functioning, track schedule adherence by product variant. A product that consistently falls behind its planned ratio signals a hidden capacity constraint or a changeover issue. The goal is not just hitting a total volume number, but hitting it in the correct proportional mix every single day.
Overtime hours serve as the ultimate diagnostic tool. The feast-or-famine overtime pattern is the clearest indicator that leveling has failed. A successful implementation stabilises the workload, effectively eliminating emergency weekend shifts. Overtime should become an exception for genuine demand spikes, not a standard operating procedure for poor planning.
Finally, monitor inventory turns and defect rates. As you produce closer to demand and reduce buffer stocks, inventory turns will increase. Simultaneously, track defect rate against batch size. You will find a direct, mathematical correlation: smaller, leveled batches consistently yield lower scrap rates and higher first-pass yield.