5. Flow, Cellular Manufacturing and Line Balancing

Introduction and Context

Lean manufacturing seeks to create value with the least possible interruption, delay, excess inventory, movement, and rework. Among the most visible expressions of lean thinking is flow: the progressive movement of materials, information, and work through a process without unnecessary waiting or accumulation. Flow is not merely a matter of arranging machines in a line. It is a managerial and technical system that connects demand, capacity, layout, work methods, quality, and production control.

Traditional functional layouts group similar machines together—for example, all lathes in one department and all milling machines in another. Such arrangements may provide high machine specialization, but they often create long transport distances, queues, large batches, and weak visibility of problems. Cellular manufacturing responds by grouping dissimilar machines and supporting activities according to product families. Line balancing then allocates work among stations so that production can meet customer demand at a controlled rhythm.

This chapter examines the design of continuous-flow and cellular systems. It develops the logic of takt time, cycle time, capacity, work content, one-piece flow, heijunka, mixed-model production, layout analysis, and balancing methods. It also considers the managerial trade-offs among utilization, flexibility, inventory, responsiveness, and resilience.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Explain the relationship among takt time, cycle time, capacity, lead time, and work-in-process inventory.

2. Distinguish functional, cellular, product-flow, and hybrid layouts.

3. Design a basic cellular manufacturing system using product-family analysis and process mapping.

4. Calculate theoretical staffing requirements and identify line imbalance.

5. Apply precedence relationships and balancing heuristics to allocate work.

6. Explain one-piece flow, heijunka, pitch, and mixed-model production.

7. Evaluate the trade-offs between utilization, flexibility, inventory, and service.

8. Recommend appropriate flow and balancing improvements in an industrial setting.

From Process Adjacency to Flow

A process flows when materials and information move through value-adding activities with minimal waiting and interruption. The ideal sequence is not simply machine A followed by machine B; it includes inspection, replenishment, setup, decision-making, and the release of accurate production information. A process can therefore have a physically attractive layout and still fail to flow if quality problems, shortages, long setups, or unreliable equipment interrupt movement.

The distinction between processing time and elapsed lead time is important. Processing time is the time during which work is actively transformed. Lead time includes processing, waiting, transport, inspection, queueing, and storage. In many factories, only a small fraction of total lead time is value-adding. Flow improvement seeks to reduce the non-value-adding portion while protecting quality and employee safety.

A useful relationship is Little’s Law:

WIP = Throughput × Lead Time

If a process produces 40 units per hour and has an average lead time of three hours, its average work-in-process is 120 units. If throughput remains constant and lead time is reduced to one hour, WIP falls to 40 units. The equation does not itself improve performance; it reveals the operational consequence of excessive queues and inventory.

Flow should be understood at several levels. Within a workstation, it concerns the sequence of motions and tasks. Across a cell, it concerns the movement of parts between processes. Across the plant, it concerns the relationship between fabrication, assembly, logistics, suppliers, and customers. Improvements at one level can create problems at another. For example, a very fast machining department may produce more components than assembly requires, increasing inventory rather than improving customer service.

Takt Time, Cycle Time and Capacity

Takt time is the rate at which a product must be completed to satisfy customer demand. It is calculated as:

Takt time = Available production time ÷ Required customer quantity

Suppose a shift provides 420 minutes of net available time after planned breaks and meetings, and demand is 210 units. Takt time is therefore 2 minutes per unit. This means that, on average, one finished unit must leave the process every two minutes. Takt is a demand signal, not a machine-speed target. It may change as demand changes.

Cycle time is the actual elapsed time between completed units at a process or station. For a stable process, cycle time should be equal to or less than takt time. If the cycle time is greater than takt, the process cannot meet demand without overtime, additional resources, inventory from an earlier period, or lost orders. If cycle time is much lower than takt, capacity exists, but the resource may be underutilized unless it supports other products or work.

The minimum theoretical number of workstations is:

Nmin = Total work content ÷ Takt time

The result must be rounded upward to a whole number. If total work content is 18 minutes and takt time is 2 minutes, the theoretical minimum is nine stations. Actual staffing may be higher because of precedence constraints, walking, ergonomic limits, product variation, absenteeism, downtime, and the need for operating flexibility.

Line efficiency can be expressed as:

Line efficiency = Total work content ÷ (Number of stations × Cycle time) × 100

Balance delay is the unused fraction of available station time:

Balance delay = 100% − Line efficiency

These measures are diagnostic rather than absolute performance judgments. A line with high efficiency may have little resilience and be highly vulnerable to disruption. A slightly less efficient line may provide better quality, safer work, and greater flexibility.

A related measure is theoretical capacity. If one station operates for 420 minutes at a two-minute cycle time, its ideal capacity is 210 units per shift. Practical capacity must account for availability, performance losses, quality losses, and planned changeovers. Overall equipment effectiveness is often used to examine these losses, but OEE should not be used to justify producing unnecessary inventory.

Cellular Manufacturing and Product Families

Cellular manufacturing arranges resources into a cell dedicated to a family of products requiring similar processing steps. A cell may contain machines, tools, inspection equipment, material presentation, and visual controls. The objective is to create a compact flow that reduces transport, queueing, handling, and coordination effort.

Product-family identification is central to cell design. Products can be grouped using routing records, process sequences, machine requirements, dimensions, materials, or tooling. A part-family matrix places parts in rows and machines or processes in columns. Similar routing patterns reveal possible families. When similarity is incomplete, management must decide whether to create a dedicated cell, a shared resource, or a hybrid arrangement.

Common layouts include:

• Functional layout: Similar machines are grouped together. It provides broad process capability but usually creates complex routings and queues.
• Product layout: Resources are arranged according to a common product sequence. It supports high-volume flow but may be inflexible for variety.
• Cellular layout: Machines for a product family are located together, often in a U-shape or compact loop.
• Hybrid layout: Different areas use different principles, such as cells for repetitive products and functional departments for specialist or low-volume work.

The U-shaped cell is useful because entry and exit can be close together, enabling one operator to monitor several processes and allowing staffing to expand or contract. However, a U-shape is not automatically lean. It must be supported by appropriate work content, safe walking distances, reliable equipment, clear standard work, and balanced material replenishment.

Cell design involves more than machine placement. Designers must consider floor space, utilities, maintenance access, changeover, quality inspection, operator visibility, material presentation, ergonomic reach, emergency access, and future demand. Excessive dedication of machines to a cell can reduce flexibility; insufficient dedication can recreate the queues of a functional layout.

One-Piece Flow and Pull Relationships

One-piece flow means that a unit moves to the next operation as soon as the required work is completed, rather than waiting for a large batch. It reduces WIP and exposes defects quickly. It also shortens feedback loops: a quality problem becomes visible after one or a few units rather than after an entire batch has been processed.

One-piece flow is most feasible when processes have compatible cycle times, stable quality, short changeovers, reliable equipment, balanced work content, and manageable product variation. It is not always economically or technically appropriate. Heat treatment, paint curing, large-scale chemical processing, and some high-speed operations may require batch processing. In such cases, the objective is to control batch size, synchronize release, and establish supermarkets or deliberate decoupling points.

A pull system authorizes production based on downstream consumption. Kanban cards, containers, electronic signals, or other replenishment mechanisms can regulate movement. A supermarket is a controlled inventory point from which downstream users withdraw material and upstream processes replenish what was consumed. Supermarkets are not a substitute for process improvement; they are a way of managing necessary decoupling and variation.

The appropriate amount of WIP depends on more than the desire to minimize inventory. Zero inventory can make a system fragile when suppliers, machines, or transport are unreliable. Excessive WIP hides problems and increases lead time. The design question is therefore: what inventory is necessary to protect customer service and process stability, and what inventory merely conceals waste?

Heijunka and Mixed-Model Production

Heijunka is production leveling by volume and mix over a defined time period. Without leveling, a plant may respond to a weekly demand total by producing large batches of one model and then changing to another. This creates peaks in material requirements, uneven workloads, shortages, overtime, and long customer lead times.

Suppose daily demand is 60 units: 30 of Model A and 30 of Model B. A leveled sequence might alternate A and B, subject to setup requirements and operational constraints. If the production interval is 10 minutes and the process must produce one finished unit every two minutes, the pitch—the time required to produce and move a standard container quantity—can be used to coordinate replenishment and management response.

Mixed-model production combines different product variants on the same line. It increases responsiveness and reduces finished-goods inventory, but it introduces variation in work content, components, tooling, and quality requirements. Successful mixed-model production requires:

• Standardized work for each variant.
• A feasible sequence that prevents excessive workload peaks.
• Point-of-use material presentation and error-proofing.
• Rapid, preferably internalized, changeovers.
• Visual identification and reliable production information.
• A method for addressing variant-specific quality checks.

Leveling should not be interpreted as producing every model in exactly equal quantities. It means distributing demand in a way that reduces avoidable instability while respecting economic, technical, and market constraints.

Line Balancing as a Design Problem

Line balancing assigns tasks to workstations so that precedence requirements are respected and the workload at each station is as close as practical to the required cycle time. The problem contains three linked elements: task times, precedence relationships, and station capacity.

A precedence diagram represents which tasks must occur before others. For example, a product may require cutting before drilling, drilling before inspection, and inspection before assembly. A task cannot be assigned merely because its duration fits; it must also be eligible according to the sequence.

A basic balancing procedure is:

1. Confirm demand, available time, and takt or target cycle time.

2. List each task, its standard time, and immediate predecessors.

3. Calculate total work content and the theoretical minimum number of stations.

4. Draw or verify the precedence diagram.

5. Assign eligible tasks to stations without exceeding cycle time.

6. Calculate efficiency, balance delay, and station utilization.

7. Test the design against ergonomics, quality, changeovers, downtime, and model variation.

8. Improve through task combination, method improvement, parallel resources, automation, or redesign.

Common heuristics include assigning the task with the largest positional weight, the longest eligible task, or the greatest number of successors. Positional weight is the time of a task plus the times of all tasks that follow it. Heuristics provide practical solutions, but they do not always produce the mathematical optimum. For complex lines, integer programming, simulation, or specialized optimization software may be justified.

Consider five tasks with times of 1.0, 1.2, 0.8, 1.5, and 0.9 minutes, giving total work content of 5.4 minutes. If the target cycle time is 1.8 minutes, the theoretical minimum is three stations. A feasible allocation might be Station 1: 1.0 + 0.8 = 1.8; Station 2: 1.2 = 1.2; Station 3: 1.5 + 0.9 = 2.4, which is infeasible. The designer must examine precedence and redistribute tasks, use four stations, improve methods, or create parallel capacity. The arithmetic minimum does not guarantee a feasible line.

Balancing must also account for variation. If task time varies around its standard, a station operating exactly at takt may frequently fall behind. Practical designs often include small capacity allowances, cross-training, and rapid support rather than maximizing every second of labor utilization.

Trade-offs in Flow-System Design

Flow design is an exercise in managing trade-offs rather than maximizing a single measure. High utilization lowers apparent unit cost when demand is stable, but it can lengthen queues and make the system slow to recover from disruptions. Flexibility requires cross-trained labor, adaptable equipment, smaller batches, and capacity margin; these may increase short-term cost while improving service and reducing finished-goods inventory.

Dedicated cells normally improve material flow and ownership but may duplicate equipment. Functional layouts share expensive resources and support product variety but often create complex scheduling and transport. One-piece flow reduces WIP but exposes every weakness in equipment reliability and process capability. Heijunka reduces demand amplification but may require frequent changeovers. Automation can improve repeatability and ergonomics, but poorly designed automation may increase downtime, capital commitment, and maintenance dependence.

Managers should evaluate at least the following outcomes together: customer lead time, on-time delivery, first-pass yield, WIP, labor productivity, safety, changeover time, equipment availability, flexibility, and total cost. A local improvement that raises machine utilization but increases WIP and delays customer orders is not operational excellence.

Illustrative Case: Redesigning a Small Appliance Assembly System

Background

The following is an illustrative case. Meridian Appliances assembles two models of countertop water purifier. Demand is 240 units per eight-hour shift, but the effective available time is 420 minutes because of breaks, meetings, and planned maintenance. The existing system uses functional departments: subassembly, wiring, testing, and packing.

Situation and Problem

The plant releases batches of 40 units. Average lead time through assembly is two shifts, even though direct processing time is approximately 12 minutes per unit. WIP frequently exceeds 500 units. Operators report that testing discovers wiring errors created many hours earlier. Model changes produce shortages of variant-specific components, and the final assembly area often works overtime.

Analysis

Takt time is 420 ÷ 240 = 1.75 minutes per finished unit. The product-family review shows that both models follow the same general sequence, with modest differences in pump and control-panel installation. A spaghetti diagram reveals extensive movement between departments. The primary causes of delay are batch transfer, unbalanced work, late defect detection, and material presentation outside the point of use.

The total work content is 12.6 minutes, producing a theoretical minimum of 12.6 ÷ 1.75 = 7.2, or eight stations. A direct eight-station design would be risky because several tasks vary by model and the electrical test has occasional downtime.

Approach and Solution

The improvement team creates a U-shaped mixed-model cell with nine operator positions, a small replenishment supermarket, and point-of-use kits for model-specific parts. Work is redistributed using a precedence diagram. The test station is positioned near the end of the cell, while a quality confirmation is placed immediately after wiring. Changeover instructions are standardized, and the two models are sequenced in a leveled pattern rather than in batches of 40.

The cell uses a pitch based on a container of five units: five units must leave approximately every 8.75 minutes. A team leader monitors the pitch board, responds to abnormalities, and stops the process when a defect cannot be corrected within the standard response time. A small finished-goods buffer remains for shipping protection, but WIP between operations is limited to clearly defined quantities.

Outcome

In the illustrative scenario, average lead time falls substantially because batches are replaced by smaller transfers. Defects are detected closer to their source, material shortages become more visible, and overtime is reduced. The additional operator position lowers theoretical labor utilization, but the system gains model flexibility and more stable delivery. The result demonstrates why efficiency percentage alone is an inadequate decision criterion.

Learning Points

The redesign succeeds because layout, balancing, material control, quality, and management response are treated as one system. A cell without leveled production would still experience variation. A balanced line without reliable material presentation would still stop. A supermarket without consumption discipline would merely become another accumulation point.

Implementation Challenges and Common Errors

Flow projects often fail when management moves machines before understanding demand, product families, task precedence, and material routes. Other common errors include designing to average demand rather than the required service level, ignoring operator fatigue, treating standard times as fixed truths, and balancing only one model while neglecting the mix.

A further error is confusing maximum utilization with maximum performance. At 100 percent planned utilization, there is no room for normal variation, minor stoppages, training, improvement activity, or unexpected demand. Another error is reducing buffers without first improving quality and reliability. Inventory may be wasteful, but removing it prematurely can shift the problem into missed deliveries.

Employees should participate in cell design because they understand actual motion, reach, interruptions, quality risks, and maintenance difficulties. Work combination must respect safety, human capability, legal requirements, and reasonable fatigue exposure. Standard work is not a mechanism for forcing unsafe speed; it is a baseline for safe, repeatable improvement.

Technology, Analytics and Emerging Practice

Digital production systems can support flow design through electronic kanban, real-time downtime visibility, digital work instructions, location tracking, and production dashboards. Simulation allows managers to test alternative layouts, staffing levels, sequencing rules, and buffer sizes before physically moving equipment. Discrete-event simulation is particularly useful when demand, downtime, or processing times are variable.

Machine-learning tools may identify patterns associated with stoppages or quality defects, but their recommendations require operational interpretation. Poor data, unstable processes, or unexamined changes in product mix can make analytics misleading. Digital systems should strengthen visual management and decision-making rather than replace direct observation at the process.

Emerging practice also emphasizes reconfigurable cells, collaborative robots, modular fixtures, additive manufacturing for tooling, and human-centered automation. These technologies are valuable when they address a clear constraint, ergonomic risk, quality issue, or flexibility requirement. Automation introduced merely to increase local speed can intensify downstream imbalance.

Best-Practice Principles for Managers

Effective flow and cell design generally follow several principles:

• Begin with customer demand and available time, not with existing machine capacity.
• Separate product-family decisions from layout preferences.
• Use takt as a planning reference and cycle time as an operational test.
• Design for the actual product mix and reasonable variation.
• Make quality conditions visible at the source.
• Keep material quantities controlled and replenish by consumption.
• Build in cross-training and manageable capacity protection.
• Treat safety, ergonomics, and maintainability as design requirements.
• Measure lead time, WIP, quality, delivery, and flexibility together.
• Rebalance whenever demand, product design, staffing, or process capability changes.

Chapter Summary

Flow is the coordinated movement of materials, information, and work with minimal waiting and interruption. Takt time translates customer demand into a required production rhythm, while cycle time describes actual process performance. The theoretical minimum number of stations is based on total work content and takt, but precedence, variation, quality, ergonomics, and reliability determine the feasible design.

Cellular manufacturing groups resources around product families to reduce transport, queues, and coordination. One-piece flow lowers WIP and accelerates feedback, although deliberate buffers and batch processes remain appropriate in some situations. Heijunka levels volume and mix, while mixed-model production requires standardized work, rapid changeovers, accurate material presentation, and robust sequencing.

Line balancing assigns tasks to stations within cycle-time and precedence constraints. Efficiency measures are useful, but operational excellence requires broader evaluation of service, flexibility, safety, quality, inventory, and resilience. Digital tools, simulation, and analytics can improve decisions when grounded in reliable data and direct process knowledge.

Key Terms

Flow: The progressive movement of work through value-adding activities with minimal interruption.

Takt time: Available production time divided by required customer demand.

Cycle time: Elapsed time between completed units at a process or station.

Work content: The total standard processing time required to complete one unit.

Line balancing: Allocation of tasks among stations to meet a target cycle time while respecting precedence.

Precedence relationship: A requirement that one task must occur before another.

Cellular manufacturing: Arrangement of dissimilar resources to process a family of similar products.

Product family: A group of products sharing similar processing or routing characteristics.

One-piece flow: Movement of units individually or in very small quantities between operations.

Work-in-process (WIP): Partially completed material within the production system.

Heijunka: Production leveling by volume and product mix.

Mixed-model production: Production of different product variants on the same line or cell.

Pitch: The time required to produce and move a defined standard quantity.

Supermarket: A controlled inventory point replenished according to downstream consumption.

Kanban: A visual or electronic signal authorizing replenishment or production.

Line efficiency: Work content divided by total available station time, expressed as a percentage.

Balance delay: The unused portion of available station time caused by imbalance.

Functional layout: Arrangement of similar machines or processes in common departments.

Standard work: The documented best-known sequence, time, and inventory required for a task.

Reconfigurable cell: A cell designed to change capacity, routing, or equipment arrangement as requirements change.

Review Questions

Short Answer Questions

1. How does takt time differ from cycle time?

2. State the relationship expressed by Little’s Law.

3. What is the purpose of a product-family matrix?

4. Define heijunka and explain its connection with production stability.

5. Why can the theoretical minimum number of stations be infeasible in practice?

Long Answer Questions

1. Explain how a cellular layout differs from a functional layout, including the conditions under which each may be appropriate.

2. Describe a systematic procedure for balancing an assembly line using task times and precedence relationships.

3. Discuss the operational benefits and limitations of one-piece flow.

4. Explain how mixed-model production and heijunka influence changeover, inventory, and customer responsiveness.

5. Critically evaluate the claim that maximum resource utilization is the primary objective of lean flow design.

Analytical / Application Questions

1. A plant has 360 net production minutes per shift and demand of 180 units. Total work content is 10.8 minutes. Calculate takt time and the theoretical minimum number of stations. Explain why the actual number may be higher.

2. A line has six stations, a target cycle time of two minutes, and total work content of 10.2 minutes. Calculate line efficiency and balance delay. Recommend two possible improvement actions.

3. A factory experiences high WIP despite excellent machine utilization. Using Little’s Law and flow principles, diagnose possible causes and propose a redesign involving layout, release control, and quality feedback.

22. Multiple Choice Questions with Answers

Question 1

A plant has 420 minutes of net available time and customer demand of 210 units per shift. What is the takt time?

A. 0.5 minute per unit

B. 2 minutes per unit

C. 210 minutes per unit

D. 630 minutes per unit

Correct Answer: B. 2 minutes per unit

Question 2

Which statement best distinguishes takt time from cycle time?

A. Takt time is based on demand, while cycle time describes actual production performance.

B. Takt time measures quality, while cycle time measures inventory.

C. Takt time is always shorter than cycle time.

D. Cycle time is calculated only for finished-goods warehouses.

Correct Answer: A. Takt time is based on demand, while cycle time describes actual production performance.

Question 3

A process has throughput of 40 units per hour and average lead time of three hours. According to Little’s Law, average WIP is:

A. 13.3 units

B. 40 units

C. 120 units

D. 360 units

Correct Answer: C. 120 units

Question 4

What is the principal purpose of a product-family analysis in cellular manufacturing?

A. To maximize the utilization of every individual machine

B. To identify products with similar processing or routing requirements

C. To eliminate all shared resources

D. To determine employee compensation

Correct Answer: B. To identify products with similar processing or routing requirements

Question 5

Total work content is 18 minutes and takt time is two minutes. What is the theoretical minimum number of stations?

A. 6

B. 8

C. 9

D. 36

Correct Answer: C. 9

Question 6

Which condition most strongly supports one-piece flow?

A. Large batch sizes and unstable quality

B. Long changeovers and unreliable equipment

C. Stable processes, compatible cycle times, and rapid feedback

D. High WIP and delayed inspection

Correct Answer: C. Stable processes, compatible cycle times, and rapid feedback

Question 7

The main purpose of heijunka is to:

A. Increase production variation between shifts

B. Level production volume and mix over time

C. Eliminate all finished-goods inventory immediately

D. Increase the size of production batches

Correct Answer: B. Level production volume and mix over time

Question 8

A line has total work content of 10.2 minutes, six stations, and a cycle time of two minutes. Its line efficiency is:

A. 51%

B. 60%

C. 85%

D. 120%

Correct Answer: C. 85%

Question 9

Which is a likely consequence of maximizing utilization at every station without capacity margin?

A. Greater resilience to disruption

B. No effect on queues

C. Greater vulnerability to variation and stoppages

D. Automatic improvement in quality

Correct Answer: C. Greater vulnerability to variation and stoppages

Question 10

In the illustrative appliance case, why was a nine-position cell selected despite a theoretical minimum of eight stations?

A. To increase batch size

B. To accommodate variation and provide practical flexibility

C. Because takt time was ignored

D. To restore the original functional layout

Correct Answer: B. To accommodate variation and provide practical flexibility