Flow, Cellular Operations and Production Leveling

Introduction

Lean manufacturing seeks to create value with the least necessary use of time, space, inventory, movement, and managerial attention. Flow is the operational expression of this purpose: materials, information, and decisions should move through a process steadily and predictably rather than accumulating in queues. Cellular operations arrange resources so that products or services follow a short, visible, and relatively uninterrupted path. Production leveling, or heijunka, stabilizes the release of work by smoothing volume and mix over time.

These ideas are interdependent. A flow line without reliable changeovers becomes dependent on large batches. A cellular layout without balanced work creates local bottlenecks. Leveling demand without sufficient capacity creates chronic lateness. Effective design therefore requires simultaneous attention to process sequence, capacity, variability, changeover, quality, material supply, and employee capability.

Learning Objectives

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

1. Explain the principles and conditions of continuous flow and one-piece flow.
2. Design and evaluate cellular layouts using product–process relationships.
3. Calculate takt time, cycle time, theoretical minimum stations, line efficiency, and balance delay.
4. Identify bottlenecks and apply the Theory of Constraints to flow improvement.
5. Explain SMED and use it to reduce batch size and improve flexibility.
6. Construct a basic heijunka schedule and assess its operational consequences.
7. Evaluate the managerial, human, technological, and governance implications of flow systems.

1. From Functional Batches to Flow

A traditional functional layout groups similar resources together: all drilling machines occupy one area, all painting equipment another, and inspection a third. Such layouts can provide flexibility for highly varied work, but they typically produce long travel distances, large queues, difficult scheduling, and delayed quality feedback. A flow-oriented layout groups resources according to the sequence required by a product family or service type.

Flow does not mean that every item must move at identical speed. It means that the process is designed to minimize unnecessary interruption and to make abnormal conditions visible. The central design question is not “How can each department stay busy?” but “How can value move through the complete system in response to customer demand?”

Push, pull, and flow

A push system authorizes work mainly from forecasts or upstream schedules. A pull system authorizes replenishment in response to downstream consumption. Flow and pull are related but not identical: a process can have physical flow while still being controlled by a forecast-driven push schedule. In lean systems, a controlled amount of work-in-process (WIP) is often authorized through kanban, a supermarket, or a carefully defined FIFO lane.

Little’s Law links WIP, throughput, and flow time:

WIP is the average work-in-process inventory, throughput is the average completion rate, and flow time is the average time an item spends in the system. Units must be consistent; for example, if throughput is 20 units per hour and flow time is 3 hours, average WIP is 60 units. The relationship assumes a stable system over the period considered: average input and output rates are approximately equal, and the measured boundaries are clearly defined. It is a descriptive relationship, not a guarantee that reducing WIP alone will improve throughput.

One-piece flow

One-piece flow transfers an item to the next operation as soon as it is complete and accepted, rather than waiting for a large batch. Its benefits include:

• rapid detection of defects;
• shorter lead time;
• lower WIP and less space;
• simpler prioritization;
• faster learning about process instability.

One-piece flow is not always economically or technically optimal. Heat treatment, curing, sterilization, transport, or expensive setup processes may require economic batches. In such cases, the objective is to reduce the batch to the smallest practical size and create controlled decoupling points.

2. Takt Time, Cycle Time, and Flow Design

Takt time expresses the pace required to satisfy demand:

Available production time excludes planned breaks, scheduled maintenance, and other deliberately unavailable periods. Both terms must use the same time unit. Takt is a demand-derived target; it is not the same as the actual cycle time of a machine.

Cycle time is the elapsed time between successive completed units at a process or station under stated operating conditions. For a stable flow system, the effective cycle time at the constraint must be no greater than takt time if demand is to be met. If cycle time is lower than takt, capacity exists, although other losses may still prevent achievement.

Worked example: basic line capacity

A line has 420 available minutes per shift and customer demand of 210 units per shift.

If the slowest station completes one unit every 2.4 minutes, the line cannot meet demand without overtime, additional capacity, productivity improvement, or a reduction in demand. Merely increasing output at faster stations will create more WIP before the slow station.

Line balancing

Suppose total work content is 18 minutes per unit and takt time is 4 minutes. The theoretical minimum number of stations is:

Thus:

The ceiling function rounds upward because a fraction of a station cannot perform the complete assigned work. Five stations are a theoretical minimum, not necessarily a feasible design; precedence constraints, ergonomics, equipment requirements, and uneven task times may require more.

Line efficiency is:

If five stations operate at a 4-minute cycle time:

Balance delay is the unused proportion:

In this example, balance delay is 10%. Efficiency should not be maximized without regard to safety, quality, fatigue, or resilience. A line balanced exactly to average demand may have no protection against normal variation.

3. Cellular Operations and Layout Analysis

A manufacturing cell is a physical and managerial arrangement in which equipment, people, materials, and information are organized around a family of products or closely related routings. The familiar U-shaped cell can reduce walking and make the beginning and end of the process visible to one operator or team, but shape alone does not create a cell.

Product families

Cell design begins with grouping products that have sufficiently similar processing requirements. Common methods include:

• Part-machine matrix: rows represent parts and columns represent machines; similarities in required machines reveal possible families.
• Production flow analysis: examines the sequence and frequency of movement between operations.
• Coding and classification: uses attributes such as geometry, material, size, and process requirements.
• Route and volume analysis: combines demand with process similarity rather than relying on either alone.

A part–machine matrix may be reordered so that dense clusters appear along the diagonal. However, exceptional parts requiring machines outside the cell must be managed explicitly through shared resources, alternate routing, or a different family assignment.

Layout alternatives

Layout decisions should include material presentation, replenishment routes, quality checkpoints, maintenance access, emergency egress, noise, lighting, and ergonomic reach. A compact layout that creates unsafe crossings or excessive operator twisting is not lean.

Flow analysis metrics

Useful measures include total travel distance, touch time, queue time, WIP, first-pass yield, changeover frequency, and end-to-end lead time. A spaghetti diagram can reveal movement waste, while a value-stream map can show information and material flow together. These tools are diagnostic; they do not replace direct observation and employee involvement.

4. Bottlenecks, Variability, and Constraints

A bottleneck is a resource whose effective capacity limits system throughput under the current operating conditions. Its capacity may be constrained by processing time, downtime, changeovers, staffing, quality losses, or material shortages. The slowest nominal machine is not always the system constraint.

The Theory of Constraints proposes five focusing steps:

1. Identify the system constraint.
2. Exploit it by ensuring that its time is used on productive, quality-conforming work.
3. Subordinate other resources to the constraint’s requirements.
4. Elevate the constraint by adding capacity or changing the process.
5. Repeat when the constraint moves.

The principle of subordination is frequently misunderstood. Non-constraint resources should not be maximized independently if doing so creates excess WIP or consumes materials needed by the constraint.

Variability amplifies queues. Sources include variable processing times, arrivals, absenteeism, machine failure, quality rework, and changing priorities. Flow design should therefore include standard work, preventive maintenance, quick response to abnormalities, cross-training, and carefully sized buffers. A small buffer before a constraint can protect throughput; uncontrolled inventory throughout the system conceals problems.

5. SMED, Batch-Size Reduction, and Flexibility

Single-Minute Exchange of Die (SMED), associated with Shigeo Shingo, is a systematic approach to reducing changeover time. “Single minute” means a changeover measured in fewer than ten minutes, not literally one minute in every application.

A changeover includes activities required to move from the last good unit of one product to the first good unit of the next. SMED distinguishes:

• Internal activities: possible only when the equipment is stopped.
• External activities: possible while the equipment is operating.

The typical improvement sequence is:

1. Observe and document the complete changeover.
2. Separate internal and external work.
3. Convert internal activities to external activities where technically possible.
4. Simplify, parallelize, and standardize the remaining internal work.
5. Establish repeatable settings and verify first-piece quality.

Examples include preparing tools in advance, using locating pins, quick clamps, standardized settings, visual aids, and parallel work by two trained people. Changeover reduction permits smaller batches. If demand for a product family is 240 units per day and the organization wishes to produce four variants daily, a simple level schedule might allocate 60 units of each variant, subject to actual demand and capacity. The sequence should also consider setup families, quality risk, material availability, and customer priority.

Batch reduction can increase setup frequency, so the net effect depends on setup time, demand, and process stability. A useful managerial comparison is not setup time alone but total lead time, WIP, responsiveness, quality exposure, and capacity consumed by changeovers.

6. Heijunka: Production Leveling

Heijunka smooths production volume and mix over a planning horizon. It does not eliminate demand variation; it decouples short-term production from every fluctuation in customer orders, provided a suitable finished-goods or intermediate buffer exists.

A leveling box or electronic heijunka board can specify the sequence and quantity to be released. For example, daily demand may be 120 units: 60 of product A and 60 of product B. A large-batch schedule might produce all A first and all B later. A leveled sequence could alternate A and B repeatedly, subject to setup feasibility. The leveled sequence reduces the time between replenishments and makes shortages visible sooner.

Leveling requires:

• reasonably stable aggregate demand;
• reliable equipment and quality;
• short and predictable changeovers;
• defined supermarkets or finished-goods buffers;
• disciplined replenishment signals;
• agreement on the time horizon and planning rules.

Heijunka is not a license to ignore genuine demand priorities or safety-critical orders. It must be integrated with sales and operations planning, available-to-promise rules, supplier capability, and capacity management.

7. Integrated Design Method

A practical design sequence is as follows:

1. Define the customer requirement, product families, process boundaries, and service promise.
2. Collect actual demand, routing, work content, downtime, setup, quality, and staffing data.
3. Calculate takt time and identify the effective constraint.
4. Map the current flow, including information, queues, rework, and material movement.
5. Group products into families and test alternative cells or flow lines.
6. Balance work using precedence relationships and realistic capacity, not nominal times alone.
7. Reduce changeovers and define the smallest practical batch.
8. Establish pull signals, FIFO limits, supermarkets, and replenishment routes.
9. Pilot the design with standard work and visual abnormality management.
10. Monitor results and adjust the system through structured problem solving.

Performance should be assessed with a balanced set of indicators: lead time, throughput, on-time delivery, first-pass yield, WIP, changeover time, labor utilization, safety events, ergonomic observations, and employee suggestions. Local utilization is not a sufficient success measure.

8. Illustrative Case: Meridian Pumps

Background

Meridian Pumps, a fictional industrial-pump manufacturer, produced three closely related pump models in a functional layout. Customer demand averaged 96 units per day. The plant operated 480 available minutes per shift, so required takt time was:

Problem

Average order lead time was 8 days, although touch time was less than 40 minutes. WIP accumulated between machining, assembly, and testing. A test rig was the effective constraint because of long setup and repeated adjustment. The plant frequently produced large batches to avoid setups, causing late delivery of lower-volume models.

Analysis

A product–machine analysis showed that 82% of volume could be processed through a common family route. Observation revealed 55-minute average changeovers at the test rig, with tools and documents gathered after the preceding batch had stopped. Operators also walked to a distant inspection area. The existing schedule maximized departmental utilization but released work faster than the test rig could absorb it.

Approach

The improvement team created a pump cell containing machining support, subassembly, final assembly, and a nearby test point. It retained a controlled external machining supply for exceptional components. SMED analysis moved document preparation, fixture inspection, and tool staging outside the test stoppage. Quick clamps and standardized test recipes reduced changeover time to 18 minutes. Work was rebalanced around a 5-minute takt, and a small FIFO lane was placed before testing. The daily schedule alternated the three models in a repeating pattern rather than producing each model in a single large batch.

Outcome

After a controlled pilot, average WIP in the family route fell by 46%, lead time fell to 3.5 days, and on-time delivery rose from 84% to 96%. First-pass yield improved because defects were detected closer to their source. The test rig remained highly utilized, but its schedule included planned maintenance and a protected small buffer. The team did not claim that all improvements resulted from layout alone; changeover reduction, quality discipline, and release control were necessary complements.

Learning points

The case demonstrates that a cell is a system design, not simply a rearrangement of machines. The constraint must be protected, batch size must be addressed, and leveling must be matched to real capacity. It also illustrates the importance of distinguishing local efficiency from end-to-end performance.

9. Limitations, Human Factors, and Governance

Flow systems can fail when management treats them as cost-cutting devices rather than learning systems. Excessive work compression may increase fatigue, musculoskeletal risk, stress, and error. Standard work should define the current best safe method while allowing workers to identify improvements. Job rotation and cross-training require adequate skill development, not superficial reassignment.

Ethically and legally, layout redesign must preserve emergency access, machine guarding, chemical controls, noise limits, working-time rules, and non-discrimination. Digital production systems may collect worker-level data. Organizations should define legitimate purposes, access controls, retention periods, and appeal processes. Algorithmic scheduling should not penalize workers for equipment failures or impossible standards without human review.

Technology can support flow through sensors, electronic kanban, digital work instructions, simulation, and predictive maintenance. Artificial intelligence may forecast demand or detect process anomalies, but it should not replace process understanding. Poor data, unstable definitions, and automated amplification of historical bias can produce misleading schedules. Human accountability remains essential for safety, quality release, and exceptional priorities.

Best-Practice Checklist

• Design around customer demand and product families.
• Measure actual effective capacity, including downtime, setup, and quality losses.
• Keep WIP deliberately limited and visible.
• Protect constraints from starvation, blockage, and defective work.
• Reduce changeover before demanding very small batches.
• Use heijunka only with appropriate buffers and reliable processes.
• Balance work for safety and sustainable staffing, not utilization alone.
• Involve operators in layout, standard work, and problem solving.
• Review both leading indicators and end-to-end outcomes.
• Treat flow as an iterative socio-technical design problem.

Chapter Summary

Flow organizes work so that value moves with minimal interruption, excess inventory, and delay. One-piece flow improves visibility and feedback but may require decoupling where technical or economic conditions demand batches. Takt time translates customer demand into a required pace, while cycle time describes actual process performance. Line balancing assigns work within precedence and capacity constraints; theoretical station counts and efficiency measures are useful but incomplete.

Cellular operations group resources around product families to shorten routes and simplify control. Effective cells depend on family formation, balanced work, material presentation, quality at the source, and safe ergonomics. Bottlenecks determine system capacity, and the Theory of Constraints emphasizes exploiting and subordinating around the constraint rather than maximizing every local resource.

SMED separates internal and external changeover work and enables smaller batches. Heijunka levels production volume and mix, reducing the disruptive effects of demand variation when supported by reliable processes, suitable buffers, and disciplined pull control. Successful implementation combines physical layout, scheduling, employee capability, technology, safety, and governance. The central managerial lesson is that end-to-end flow—not isolated utilization—is the appropriate unit of improvement.

Key Terms

Takt time: The pace of production required to meet customer demand.

Cycle time: The elapsed time between successive completed units at a process.

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

Work-in-process (WIP): Items currently being processed or waiting within a system.

Little’s Law: The relationship WIP = throughput × flow time under stable conditions.

Cellular manufacturing: Arrangement of resources around a product family or common route.

Product family: Items sharing sufficiently similar processing requirements.

Line balancing: Allocation of tasks among stations to meet a required pace.

Precedence constraint: A requirement that one task occur before another.

Bottleneck: A resource limiting system throughput under current conditions.

Theory of Constraints: A method focused on identifying and managing the system constraint.

SMED: A changeover-reduction approach based on separating internal and external activities.

Changeover: Work required to move from the last good unit of one product to the first good unit of another.

Batch size: Quantity produced between successive changeovers.

Heijunka: Production leveling by volume and product mix.

Kanban: A signal authorizing replenishment or movement in a pull system.

FIFO lane: A controlled queue in which items are processed in arrival order.

Effective capacity: Practical output capacity after losses such as downtime, setup, and quality problems.

Standard work: Documented current best method for performing a task safely and consistently.

Review Questions

Short Answer Questions

1. Distinguish takt time from cycle time.
2. State the assumptions under which Little’s Law is applied.
3. What is the purpose of a product–machine matrix?
4. Define an internal changeover activity.
5. Why can maximizing local utilization damage overall flow?

Long Answer Questions

1. Explain how a functional layout differs from a cellular layout and evaluate the conditions favoring each.
2. Describe the five focusing steps of the Theory of Constraints and relate them to a flow line.
3. Explain the SMED method and discuss how setup reduction affects batch size and responsiveness.
4. Discuss the role of heijunka in managing demand variability, including its prerequisites and limitations.
5. Develop an integrated method for designing a flow cell from demand analysis through pilot implementation.

Analytical / Application Questions

1. A process has 360 available minutes per shift and demand of 90 units. Calculate takt time. If total work content is 13 minutes and the proposed line has four stations operating at takt, calculate theoretical minimum stations and line efficiency for the proposed design.
2. A constraint completes 80 units per shift while upstream departments release 120 units. Diagnose the likely flow consequences and propose actions using the Theory of Constraints.
3. A plant produces four variants in weekly batches because changeover takes 75 minutes. Design a data-collection and SMED intervention plan, and explain what evidence would justify moving toward a leveled mixed-model schedule.

Multiple Choice Questions with Answers

Question 1

A line has 420 available minutes per shift and demand is 210 units. What is the takt time?

A. 1 minute per unit

B. 2 minutes per unit

C. 3 minutes per unit

D. 4 minutes per unit

Correct Answer: B. 2 minutes per unit

Question 2

Which statement best describes one-piece flow?

A. All products are completed in one department before moving onward

B. Items move individually or in very small quantities as soon as they are accepted

C. Production is released only from a long-range forecast

D. Inventory is maximized before every operation

Correct Answer: B. Items move individually or in very small quantities as soon as they are accepted

Question 3

If total work content is 18 minutes and takt time is 4 minutes, what is the theoretical minimum number of stations?

A. 4

B. 4.5

C. 5

D. 6

Correct Answer: C. 5

Question 4

Which activity is classified as external in SMED?

A. Removing a die after the machine stops

B. Adjusting the machine while stopped

C. Testing the first unit after setup

D. Preparing tools while the previous product is still running

Correct Answer: D. Preparing tools while the previous product is still running

Question 5

What is the primary purpose of heijunka?

A. To maximize batch size

B. To level production volume and mix over time

C. To eliminate all finished-goods inventory

D. To schedule every operation independently

Correct Answer: B. To level production volume and mix over time

Question 6

Under stable conditions, throughput is 20 units per hour and flow time is 3 hours. What is average WIP?

A. 6 units

B. 17 units

C. 23 units

D. 60 units

Correct Answer: D. 60 units

Question 7

Which action represents exploitation of a constraint?

A. Allowing the constraint to wait for materials

B. Ensuring the constraint works on productive, conforming output

C. Increasing inventory at every process

D. Balancing all departments to equal utilization

Correct Answer: B. Ensuring the constraint works on productive, conforming output

Question 8

What is a principal advantage of a cellular layout?

A. It always provides unlimited routing flexibility

B. It eliminates the need for product families

C. It can shorten travel and make flow more visible

D. It requires no balancing or standard work

Correct Answer: C. It can shorten travel and make flow more visible

Question 9

A line has 18 minutes of work content, five stations, and a 4-minute cycle time. What is its line efficiency?

A. 72%

B. 80%

C. 90%

D. 100%

Correct Answer: C. 90%

Question 10

Why may a small buffer be placed before a bottleneck?

A. To protect the constraint from starvation

B. To hide all quality problems permanently

C. To ensure every upstream machine runs continuously

D. To replace preventive maintenance

Correct Answer: A. To protect the constraint from starvation