Constant Work-in-Process (CONWIP)

1. What Is Constant Work-in-Process (CONWIP)?

Constant Work-in-Process (CONWIP) is an operational production-control framework that limits how much work can enter a defined manufacturing flow. When a job leaves that flow, it frees an authorization for another eligible job to enter. Its purpose is to control congestion while sustaining required output—not simply to minimize inventory.

For builders of industrial machinery, CONWIP structures a practical question: how many machining lots, subassemblies, or configured machines should be active simultaneously? It gives executives and operations consultants a basis for redesigning order-release discipline rather than assuming that starting more orders will improve delivery.

2. Origin and Background

Mark L. Spearman, David L. Woodruff, and Wallace J. Hopp developed CONWIP at Northwestern University in the late 1980s. Their 1989 publication described a hierarchical control architecture; their influential 1990 paper, “CONWIP: a pull alternative to kanban,” appeared in the International Journal of Production Research. Thus, 1990 marks the familiar formulation, not the earliest published work.

The design sought to retain the benefits of bounded work-in-process without requiring product-specific replenishment signals throughout production. Subsequent Factory Physics work helped establish CONWIP in operations teaching and industrial practice. Its original context was manufacturing, so applying it to equipment production principally requires adapting boundaries, workload definitions, and release rules—not changing its underlying purpose.

3. How Constant Work-in-Process (CONWIP) Works

The name compresses CONstant Work In Process:

• CON — Constant: A fixed authorization ceiling for a defined operating period.
• W — Work: The specified jobs, units, or lots being controlled.
• I — In: Work admitted inside the control boundary.
• P — Process: The operations between release and the defined exit.

Establish a production loop with a clear entry and exit. Each released job receives a physical card or equivalent electronic authorization. That authorization stays occupied while the job is inside the loop. On exit, it becomes available for another job. “Constant” therefore describes the control limit; it does not require filling every slot when there is no eligible demand.

A separate backlog determines which job enters next. CONWIP cards are not permanently assigned to individual part numbers. Unlike station-level kanban, the basic system bounds total work across the loop rather than inventory at every workstation. Internal sequencing still needs dispatching rules that account for due dates and changeovers.

In equipment production, one authorization might represent a standard machine assembly or a fixed-size machining lot. However, ten small assemblies may impose very different demands from ten large ones. Where processing requirements vary materially, consider separate comparable product-family loops or a workload-based CONWIP variant, which limits defined processing hours rather than job count. This is an extension, not the original counting rule.

4. When to Use Constant Work-in-Process (CONWIP)

CONWIP is particularly useful when premature releases create long queues, frequent expediting, and uncertain manufacturing lead times. High-product-mix research demonstrates its potential to regulate total WIP, although results depend on the accompanying sequencing rules and production environment. It should not be treated as a guaranteed improvement over every alternative.

For electrical equipment, a useful application could be a configured-panel assembly and test loop. The release gate should check that engineering information and necessary components are available; an empty authorization is not sufficient reason to start an incomplete panel.

A single plant-wide cap is less attractive when routes diverge substantially or workload concentrates unpredictably at different resources. Local congestion can persist despite acceptable total WIP. Nor does a cap create capacity: persistently excessive demand still requires capacity, commercial, or delivery-date decisions.

Require traceable jobs, meaningful routing groups, and feasible output targets. Bring together production control, supervisors, quality, maintenance, and procurement. Initial design can be workshop-based, but validation should cover representative product mixes and multiple complete production cycles—not an arbitrary calendar deadline.

5. How to Apply Constant Work-in-Process (CONWIP): Step-by-Step

1. Clarify the decision and scope.

   Define whether the objective is shorter manufacturing lead time, more dependable delivery, or lower WIP while protecting output. Specify the plant, product family, entry point, exit condition, planning horizon, and responsibilities.

2. Gather operational evidence.

   Collect release and completion timestamps, routings, processing times, lot sizes, changeovers, downtime, rework, due dates, and kit readiness. Reconcile system records with physical work before using them to establish a baseline.

3. Define comparable control units.

   Choose units, fixed-size lots, or jobs with reasonably comparable processing requirements. If adopting a workload-based variant, document precisely how workload is calculated and released; do not mix job counts and processing hours within one cap.

4. Build the control artifact.

   Create a loop map, authorization register, sequenced backlog, and release board. Define treatment of inspection holds, rework, split lots, and subcontracted operations. Make the occupied and available authorizations visible to the release owner.

5. Test candidate WIP limits.

   Use historical replay, simulation, or a controlled pilot to compare caps. Seek a limit that protects required throughput without unnecessary queues. If existing WIP exceeds the chosen cap, pause new releases and allow work to drain rather than relabeling it. Cap selection and cap adjustment are distinct design decisions.

6. Translate the design into release controls.

   Update production planning and scheduling so planned dates propose jobs, while the release gate checks authorization availability, material readiness, and priority. Assign approval rights for exceptions and record every override.

7. Test sensitivities.

   Challenge the design with heavier product variants, longer setups, equipment outages, missing components, and increased rework. Include competing demand on shared test benches or machining centers. Distinguish insufficient WIP from insufficient capacity.

8. Align stakeholders and iterate.

   Review throughput, manufacturing lead time, backlog age, due-date performance, and bottleneck starvation together. Agree corrective actions with frontline teams. Change the cap deliberately, with documented evidence, rather than whenever an urgent order arrives.

6. Example: Constant Work-in-Process (CONWIP) in Action

Fictional example: Alder Hydraulic Systems builds configured hydraulic power units. Its selected product family follows assembly, wiring, and pressure testing, with one test bench serving the variants. Eighteen units are active, yet supervisors repeatedly expedite orders and move incomplete assemblies aside.

The team chooses CONWIP because it needs to control starts across this complete flow. It maps the loop from kit release through accepted pressure testing and uses one unit per authorization. Highly customized units with substantially different work content remain outside the pilot and receive separate planning treatment.

Using recorded processing times, kit availability, and test interruptions, the team compares illustrative caps of eight, twelve, and sixteen units. In this fictional analysis, eight causes frequent test-bench starvation; twelve supports planned output; sixteen adds queues without meaningful output improvement. These numbers are scenario assumptions, not industry benchmarks.

The plant adopts twelve authorizations and lets existing WIP drain. A failed pressure test retains its authorization until rework and acceptance are complete. Returned authorizations release the next ready, appropriately prioritized unit—not whichever order attracts the loudest escalation.

The subsequent manufacturing operations work aligns kit preparation, daily release reviews, supervisor routines, and performance measures. Management also monitors waiting orders outside the loop to ensure shorter shop-floor lead time does not merely conceal a growing customer backlog.

7. Strengths and Limitations

Strengths

• Visible release discipline: A bounded authorization population makes the control rule explicit and limits uncontrolled WIP growth.
• Flexibility across products: A shared authorization pool avoids maintaining separate WIP allocations for every part number, while the backlog supports deliberate sequencing.
• Executive clarity: The design separates three management questions: which orders matter, which are ready, and how much work the flow should contain.

Limitations

A job-count cap controls workload poorly when jobs differ substantially in processing requirements. Workload-based extensions address this issue but require reliable standards and explicit accounting rules. Greater product variety does not automatically make a simple counting system appropriate.

Aggregate WIP control also does not guarantee balanced local queues. Research comparing CONWIP and POLCA illustrates circumstances where more localized control can perform better. Equipment assembly therefore requires particular attention to shared resources, converging subassemblies, and route differences.

8. Common Pitfalls and How to Avoid Them

• Gaming the boundary: Moving delayed jobs outside the reported loop can improve the dashboard without improving delivery. Track total customer lead time and unreleased backlog alongside loop performance.
• Counting unequal jobs equally: Large assemblies and small replacement modules may occupy identical cards but require different effort. Segment comparable work or test a workload-based extension.
• Allowing routine overrides: Unrecorded releases undermine the cap. Establish explicit exception ownership and review their frequency before increasing authorizations.
• Rewarding local busyness: Do not release unnecessary work simply to keep every machine occupied. Evaluate completed output and delivery performance, not utilization alone.
• Reducing WIP before fixing readiness: Test whether starvation comes from the cap, missing kits, or unreliable equipment. Address the actual cause before adjusting the limit.
• Mishandling rework and scrap: Retain authorization for work still inside the loop. On genuine scrap exit, reconcile it once; any replacement job must obtain its own authorization.

9. How Constant Work-in-Process (CONWIP) Relates to Other Frameworks

Value Stream Mapping usually comes first. It exposes material and information flows and helps identify a coherent product-family boundary. CONWIP then supplies a release-control mechanism for the selected flow; it does not replace the broader diagnosis or future-state design.

Kanban is an alternative or complementary control. Product-specific replenishment loops can suit repeat components, while CONWIP permits explicit backlog sequencing without maintaining WIP for each part number. A practical equipment plant may use kanban for standard components feeding a CONWIP-controlled assembly flow.

Workload Control and POLCA deserve consideration when routing complexity dominates. Workload Control evaluates released load against resource capacity; POLCA coordinates overlapping cell-to-cell loops. Compared with basic CONWIP, these approaches place greater emphasis on workload distribution or local capacity availability. Assess them before choosing one global cap for a complex job shop.

10. Key Takeaways

• CONWIP controls entry into a defined production flow through a bounded authorization population.
• Choose boundaries and control units that reflect actual equipment-production work.
• Separate authorization availability from job priority and material readiness.
• Test WIP limits against required output, product mix, and disruption scenarios.
• Monitor unreleased backlog and local queues; do not mistake a cleaner shop floor for better customer delivery.

11. FAQs About Constant Work-in-Process (CONWIP)

How should a manufacturer choose the CONWIP limit?

Compare candidate limits against required throughput and lead-time performance under representative conditions. Avoid selecting an arbitrary inventory-reduction percentage. Processing variability, available capacity, and release timing influence the appropriate setting.

Is CONWIP better than kanban for industrial equipment?

Not universally. CONWIP supports explicit backlog sequencing and avoids part-number-specific WIP allocations. Kanban can remain useful for repeat components. The appropriate choice depends on the flow being controlled, not the industry label alone.

Can CONWIP coexist with material requirements planning?

Yes, provided their roles are separated. Use material requirements planning for requirements and proposed orders, then apply CONWIP as an additional shop-floor release gate. Planned dates should not automatically override the WIP limit.

Can a smaller manufacturer start without new software?

Start with one clearly bounded product family, numbered physical authorizations, a ready-job backlog, and daily reconciliation. Establish dependable release and completion practices first. Consider electronic integration only after the pilot’s rules are understood and consistently followed.