Scheduling, Workflow and Workforce Operations

Introduction and context

Scheduling is the operational process of deciding what work will be done, in what sequence, by which resources, and at what time. It converts plans and customer requirements into executable activities. In manufacturing, scheduling determines the release and movement of jobs through machines, labour, tools and inspection points. In services, it allocates employees, rooms, equipment and customer appointments across time.

Scheduling is closely connected with workflow—the physical and informational movement of work—and workforce operations—the planning, deployment and control of people. Poor scheduling produces queues, excessive work-in-process, missed due dates, overtime, idle resources and quality failures. Effective scheduling supports the quality-management objectives of flow stability, process capability, customer satisfaction and continual improvement.

A schedule is not merely a timetable. It is a decision system that must reconcile demand, capacity, priorities, constraints, uncertainty and performance objectives. This chapter develops practical methods for loading, sequencing, dispatching, Gantt-chart construction, line balancing, finite-capacity scheduling and workforce planning.

Learning objectives

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

1. Explain the relationship between loading, sequencing, dispatching and workflow control.
2. Construct and interpret Gantt charts for manufacturing and service operations.
3. Apply common priority rules and evaluate their consequences.
4. Calculate takt time, cycle time and line-balance efficiency.
5. Distinguish infinite-capacity from finite-capacity scheduling.
6. Develop a basic workforce plan using demand, availability and skill constraints.
7. Analyse a scheduling problem using operational, quality and ethical criteria.
8. Recommend improvements using digital scheduling and real-time control systems.

1. The architecture of operational scheduling

Scheduling decisions occur at several time horizons:

Four related activities should be distinguished.

Loading assigns work to work centres or employees. It answers: “How much work is assigned to each resource?” A load may be expressed in hours, units, appointments or cases.

Sequencing determines the order in which waiting jobs are processed. It answers: “Which job should be processed next?”

Scheduling specifies start and completion times, resource assignments and precedence relationships.

Dispatching releases or authorises the next activity at the point of execution. Dispatching is therefore the operational link between a schedule and actual workflow.

A useful workflow logic is:

Demand and orders → capacity assessment → loading → sequencing → time scheduling → dispatching → monitoring → rescheduling.

The sequence is iterative. A machine breakdown, absenteeism, material shortage or urgent customer order can make the original schedule infeasible.

Important terminology

• Job: A customer order, batch, patient, case or service request.
• Operation: A defined activity required to complete a job.
• Work centre: A machine, department, team or service resource.
• Routing: The prescribed sequence of operations for a job.
• Processing time: Time during which a resource actively works on a job.
• Setup time: Time required to prepare a resource between jobs.
• Flow time: Time from a job’s release to its completion.
• Due date: Required completion date or time.
• Makespan: Time at which the final job in a schedule is completed.
• Bottleneck: A capacity-constrained resource limiting system output.
• Work in process (WIP): Jobs started but not yet completed.
• Slack: Remaining time before a job becomes late, after considering required processing.

2. Loading and capacity analysis

Loading can be evaluated at a work-centre, shift, department or system level. The basic load in hours is:

where L is total required load in hours, qᵢ is the quantity of job i, pᵢ is processing time per unit, and sᵢ is setup time for job i. The formula assumes that processing and setup times are known and that all terms use the same time unit.

If effective available capacity is C, the load ratio is:

Here ρ is utilisation or load ratio. If ρ > 1, the planned work exceeds available capacity and cannot be completed without overtime, subcontracting, additional resources, reduced work content or a later completion date. A ratio below 1 does not guarantee good performance because variability, maintenance, breaks and quality losses consume capacity.

Effective capacity may be estimated as:

where N is the number of equivalent resources, H is scheduled hours per resource, A is availability, P is performance efficiency and Q is the proportion conforming to specification. These factors are proportions between 0 and 1. The formula is a planning approximation, not a physical law; the factors should not be multiplied if they overlap.

For example, a cell has two machines, each scheduled for 8 hours. Availability is 0.90, performance is 0.95 and first-pass quality is 0.98. Effective capacity is:

C = 2 × 8 × 0.90 × 0.95 × 0.98 = 13.41 hours of equivalent conforming production. If the planned load is 12 hours, the nominal load ratio is 12 ÷ 13.41 ≈ 0.895, or 89.5%.

High utilisation can be attractive financially but dangerous operationally. As utilisation approaches 100%, small disruptions produce rapidly increasing queues. Managers should therefore protect capacity at bottlenecks and distinguish productive utilisation from merely keeping resources busy.

3. Sequencing and priority rules

A sequencing rule determines which waiting job is selected when a resource becomes available. No single rule is best for every objective.

The critical ratio is:

where d is the due date, t is the current time and r is remaining processing time. A value below 1 indicates that the job is behind its required pace, assuming no further disruption. The rule is dynamic and should be recalculated when conditions change.

For a completed job, lateness is:

where Cᵢ is completion time and dᵢ is due date. Tardiness cannot be negative:

Average flow time is:

where rᵢ is release time and n is number of jobs. If all jobs are released at time zero, average flow time is simply the average completion time. These measures assume that the time origin and units are consistent.

Worked sequencing example

Three jobs are available at time zero:

Under SPT, the order is B, C, A. Completion times are 2, 5 and 9. Tardiness values are 0, 0 and 0. Average completion time is (2 + 5 + 9) ÷ 3 = 5.33 hours.

Under EDD, the order is B, C, A because due dates are 5, 7 and 10; therefore the result is the same. If A had a due date of 3, EDD would place A first, whereas SPT would still place B first. The appropriate decision would depend on whether the operation prioritises due-date performance or average flow time.

Priority rules must not override safety, regulatory, contractual or quality requirements. For example, a job awaiting a required inspection cannot be dispatched merely because it has a high priority.

4. Gantt charts and visual workflow control

A Gantt chart displays activities against time. Horizontal bars represent processing, setup, waiting, maintenance or downtime. A manufacturing Gantt chart can show jobs across machines; a service Gantt chart can show appointments across employees, rooms or teams.

A useful Gantt chart should identify:

• resource or work centre;
• job and operation;
• planned start and finish;
• setup and processing segments;
• planned maintenance and non-working periods;
• actual progress and deviations;
• constraint or dependency information.

For a simple single-machine schedule, start and completion times are calculated sequentially. If jobs are released at time zero and no idle time is required, the completion time of job i is the cumulative processing time of all preceding jobs and itself:

This special case assumes one resource, no setup time, no interruptions and a fixed sequence. In practice, setup, release dates, breaks and maintenance must be inserted explicitly.

Gantt charts are valuable because they make hidden waiting and overload visible. However, they are not optimisation models by themselves. A visually balanced chart may still violate material availability, employee skills or precedence constraints. The schedule must therefore be validated against the actual operating system.

5. Workflow, queues and control points

Workflow design concerns the movement of work through a process. A stable workflow normally has:

1. clear entry criteria;
2. defined process ownership;
3. standard work and routing;
4. controlled WIP limits;
5. visible status information;
6. quality checks at appropriate points; and
7. escalation rules for abnormal conditions.

Push systems release work according to a plan or forecast. Pull systems release work in response to downstream consumption or available capacity. Kanban is a common pull-control method, although a pull signal does not eliminate the need for capacity planning.

Little’s Law provides a useful relationship for a stable process:

where WIP is average work in process, λ is average throughput rate, and T is average flow time. Units must be compatible—for example, jobs per day multiplied by days gives jobs. The relationship applies to a stable system over an appropriate observation period, where average input and output rates are approximately equal. It does not by itself explain causality or identify the bottleneck.

For example, if a service team completes 40 cases per day and average case flow time is 0.75 day, average WIP is 40 × 0.75 = 30 cases. If management wants to reduce WIP without reducing throughput, it must reduce flow time through better process design, faster information transfer or removal of waiting.

6. Line balancing

Line balancing assigns tasks among workstations so that precedence relationships are respected and workloads are reasonably equal. It is especially important in repetitive manufacturing, assembly, food processing and high-volume service operations.

Takt time expresses the pace required to meet customer demand:

where tₜ is takt time per unit, Tₐ is available production time in the planning period, and D is required customer demand in units during that period. Available time should exclude planned breaks and other non-production periods. Takt time is a demand rate, not necessarily the actual processing time.

For a feasible line, each workstation’s cycle time should generally be no greater than takt time. If total task time is Σt, the theoretical minimum number of stations is:

where C is selected cycle time and ceil means round upward to the next whole station. Line-balance efficiency is:

where N is the actual number of stations. The balance delay is 100% − E. These formulas assume comparable station availability and a paced line; they do not capture walking, variability, ergonomic limits or quality losses unless those are incorporated into task times.

Worked line-balancing example

A product has five tasks with times 2, 3, 4, 2 and 3 minutes. Total task time is 14 minutes. Daily available time is 420 minutes and demand is 100 units, so takt time is 420 ÷ 100 = 4.2 minutes per unit.

The theoretical minimum number of stations is ceil(14 ÷ 4.2) = 4. Suppose precedence constraints allow the tasks to be assigned to four stations with loads 4, 3, 4 and 3 minutes. Efficiency is:

E = 14 ÷ (4 × 4.2) × 100% ≈ 83.3%.

The line is feasible on cycle time because no station exceeds 4.2 minutes. Nevertheless, the organisation should validate operator fatigue, quality checkpoints, material presentation and actual downtime before implementation.

7. Finite-capacity scheduling

Infinite-capacity scheduling plans work as if every resource can accept unlimited load. It is useful for rough planning but may generate impossible schedules. Finite-capacity scheduling recognises actual limits on machines, people, rooms, tools, shifts, materials and calendars.

A finite-capacity schedule must respect:

• resource capacity and working calendars;
• job release dates;
• operation precedence;
• setup and changeover requirements;
• material and tooling availability;
• employee skills and legal working limits;
• maintenance and planned downtime;
• quality holds and inspection requirements.

A simple capacity-feasibility condition for a resource over a period is:

This condition is necessary but may not be sufficient because work may be concentrated in a particular time interval or require a sequence that creates additional setups.

Finite scheduling can be performed manually for small systems, with spreadsheet models, or with advanced planning and scheduling software. Common computational approaches include dispatching heuristics, bottleneck scheduling, mixed-integer optimisation, constraint programming and simulation. In practice, a good feasible schedule is often more valuable than an impressive but unstable theoretical optimum.

The bottleneck should receive special attention. The five focusing steps of the Theory of Constraints provide a practical logic: identify the constraint, exploit it, subordinate other resources to it, elevate its capacity, and repeat when the constraint moves. Scheduling the bottleneck continuously is not always correct; excessive WIP before it can increase lead time and conceal quality problems.

8. Workforce planning and workforce operations

Workforce planning translates expected workload into staffing, skills, shifts and deployment. It must address both quantity and capability. A department may have enough employees in total but lack the required certified skill at the required time.

A basic workforce requirement is:

where R is required full-time-equivalent workforce, W is workload hours, and hₑ is effective productive hours per employee in the planning period. Effective hours should account for breaks, training, meetings, leave and expected absence. The result must be rounded upward where partial staffing is not operationally possible.

A skills matrix records which employees are trained and authorised for each task. Workforce plans should also include:

• shift patterns and handover time;
• overtime and rest requirements;
• cross-training and succession coverage;
• temporary or agency labour;
• ergonomic and psychosocial risks;
• absence and turnover assumptions;
• equal-opportunity and non-discrimination controls.

A fair schedule is not necessarily identical for every employee. It should be transparent, consistent with legitimate operational needs, attentive to reasonable accommodation and compliant with employment law and collective agreements. Algorithmic scheduling systems require governance because historical data may reproduce unfair allocation of desirable shifts or excessive working hours.

In service environments, demand may vary by hour. Workforce planning therefore often combines forecasts with queue information. Examples include nurses by ward, call-centre agents by interval, technicians by territory and lecturers by class timetable. Managers should avoid treating employees as interchangeable units when competence, experience and emotional demands differ.

9. Technology, analytics and AI

Modern scheduling systems integrate enterprise-resource-planning data, manufacturing-execution systems, warehouse systems, time-and-attendance records and customer appointment platforms. Real-time dashboards can display schedule adherence, WIP, bottleneck status, absence and quality alerts.

AI and machine learning can support demand forecasting, estimated processing times, predictive maintenance and schedule recommendations. However, an AI-generated schedule is only as reliable as its data, assumptions and constraints. Governance should include:

• human review before consequential decisions;
• explainable priority and allocation logic;
• protection of personal and operational data;
• monitoring for bias and systematic disadvantage;
• version control and audit trails;
• fallback procedures when systems fail.

Digital tools should support standardised decision-making, not replace operational knowledge. A schedule that optimises utilisation while increasing defects, unsafe pace or employee fatigue is not a quality schedule.

10. Illustrative case: scheduling a medical-device assembly cell

Background

A medical-device manufacturer assembles three product families on one cell. Demand has increased, but late orders and overtime have also increased. The cell contains inspection, assembly and packaging activities. Two operators are qualified for inspection, four for assembly and three for packaging.

Problem

The planner uses infinite-capacity spreadsheets. The resulting plan assigns 46 labour-hours to a shift with only 40 effective labour-hours. Changeovers are omitted, and urgent orders are inserted informally. Operators report frequent interruptions and inspection queues.

Analysis

The team first separates processing, setup and inspection time. It then constructs a skills matrix and identifies inspection as the constraint. A finite-capacity plan is created using due-date priority, but no job can bypass mandatory inspection. A Gantt chart shows that assembly is often completed faster than inspection can accept it, creating WIP.

The team calculates demand pace. The shift provides 420 available minutes and demand is 100 equivalent units, giving a takt time of 4.2 minutes per unit. Inspection requires 5 minutes per equivalent unit, so one inspector cannot meet demand. The organisation must add qualified inspection capacity, reduce inspection work content, improve first-pass quality or revise demand and release decisions.

Approach and solution

Management cross-trains one assembly operator in inspection, introduces a controlled qualification process, and schedules inspection continuously rather than in batches. A small WIP limit is placed before inspection. The dispatching rule is earliest due date, with a critical-ratio review for disrupted orders. Setup times are included in the finite-capacity schedule, and a daily 15-minute schedule review is established.

Outcome

In this illustrative case, the revised plan reduces unplanned overtime and prevents assembly from releasing work that inspection cannot absorb. The principal improvement is not simply a faster schedule; it is alignment between demand, qualified capacity, release control and quality gates.

Learning points

1. Capacity must be measured at the bottleneck, not only as a department average.
2. Quality inspection is part of capacity, not an administrative afterthought.
3. Cross-training increases flexibility but requires competence assurance.
4. WIP limits expose constraints and reduce hidden queues.
5. A schedule should be reviewed using both delivery and quality measures.

11. Performance measurement and continual improvement

Scheduling performance should be evaluated using a balanced set of measures:

Schedule adherence compares actual progress with the approved schedule, but it should not encourage employees to conceal defects or unsafe conditions. Root-cause analysis should distinguish inaccurate standards, poor release control, unreliable equipment, absenteeism, demand volatility and inadequate skills.

A practical improvement cycle is: establish a stable baseline, identify the dominant constraint, test a controlled change, measure delivery and quality together, standardise successful practice, and revise the schedule rules when process conditions change.

Challenges and limitations

Scheduling models simplify reality. Processing times may be uncertain; jobs may be reworked; priorities may conflict; employees may need breaks; and customers may change requirements. Optimising one measure can damage another: maximum utilisation may increase lead time, minimum labour cost may reduce resilience, and strict adherence may prevent sensible response to emergencies.

Managers should therefore use scenario analysis and sensitivity testing. Ask what happens if demand rises, a skilled employee is absent, a machine loses two hours, or first-pass yield falls. Robust schedules include practical buffers, escalation rules and authority to stop or reschedule work.

Ethically, scheduling must protect health, safety, dignity, privacy and fair treatment. Regulatory requirements may govern working hours, medical staffing, product traceability, competence, rest periods and quality release. In regulated operations, unauthorised schedule changes can compromise compliance and product safety.

Chapter Summary

Scheduling converts demand into executable work assignments and time decisions. Loading measures required work against available capacity; sequencing chooses the order of competing jobs; scheduling sets dates and resources; and dispatching releases work at the operating point. Gantt charts make timing, dependencies and overload visible, while workflow controls manage WIP and process stability.

Priority rules such as FCFS, SPT, EDD and critical ratio serve different objectives and must be selected deliberately. Takt time and line-balancing measures help match repetitive operations to demand. Finite-capacity scheduling improves realism by recognising resource, skill, material, maintenance and quality constraints. Workforce planning must consider effective hours, competence, shifts, fairness and legal requirements. Digital and AI tools can improve responsiveness, but they require reliable data, human oversight and governance. The central quality-management principle is that delivery performance, capacity utilisation, workforce wellbeing and conformance must be managed together.

Key Terms

Loading: Assigning work requirements to resources.

Sequencing: Determining the order in which jobs are processed.

Dispatching: Authorising the next job or operation for execution.

Gantt chart: A time-scaled visual display of activities and resources.

Work centre: A machine, team, department or service resource.

Routing: The ordered path of operations for a job.

Processing time: Active time spent performing an operation.

Setup time: Time required to prepare a resource for a job.

Flow time: Elapsed time from job release to completion.

Makespan: Completion time of the final job in a schedule.

Bottleneck: Resource that constrains system capacity.

Work in process: Started but incomplete work.

Takt time: Available production time divided by required demand.

Cycle time: Time between successive completed units or processing time permitted per unit.

Line balancing: Allocating tasks to stations to meet pace and reduce imbalance.

Finite-capacity scheduling: Scheduling that respects actual resource limits.

Critical ratio: Time remaining divided by work remaining.

Skills matrix: Record of employee competence and authorisation by task.

Schedule adherence: Degree to which actual execution follows the approved schedule.

Little’s Law: Stable-system relationship connecting WIP, throughput and flow time.

Review Questions

Short Answer Questions

1. Distinguish loading, sequencing and dispatching.
2. What information does a Gantt chart provide?
3. Why can a load ratio below 100% still produce queues?
4. Define takt time and explain how it differs from cycle time.
5. State two governance risks associated with AI-based workforce scheduling.

Long Answer Questions

1. Explain how a manufacturing organisation should move from demand information to a dispatching decision.
2. Compare FCFS, SPT, EDD and critical-ratio sequencing rules, including one advantage and limitation of each.
3. Explain finite-capacity scheduling and discuss why infinite-capacity plans may fail in practice.
4. Describe the role of line balancing in a repetitive production or service process, including relevant calculations.
5. Develop a workforce-planning approach for a service operation with variable hourly demand.

Analytical / Application Questions

1. A work centre has 10 available hours and receives three jobs requiring 4, 3 and 5 hours respectively. Analyse the capacity problem and recommend at least two responses while considering quality and employee wellbeing.
2. A line has 480 available minutes per day and demand of 120 units. Four tasks require 1.5, 1.0, 2.0 and 1.5 minutes, with precedence constraints requiring the first two tasks before the final two. Calculate takt time, theoretical minimum stations and the efficiency of a feasible two-station assignment if each station has 3-minute cycle time.
3. A call centre experiences high utilisation but worsening customer waiting time and employee absence. Use workflow, queue and workforce concepts to diagnose the likely causes and propose a controlled improvement plan.

Multiple Choice Questions with Answers

Question 1

Which activity determines the order in which waiting jobs are processed?

A. Loading

B. Sequencing

C. Auditing

D. Forecasting

Correct Answer: B. Sequencing

Question 2

A resource has a required load of 18 hours and effective capacity of 20 hours. Its load ratio is:

A. 0.90

B. 1.11

C. 2.00

D. 38.00

Correct Answer: A. 0.90

Question 3

Which priority rule generally places the job with the earliest due date first?

A. SPT

B. LPT

C. EDD

D. FCFS

Correct Answer: C. EDD

Question 4

A process completes 50 jobs per day and has average flow time of 0.4 day. According to Little’s Law, average WIP is:

A. 20 jobs

B. 50 jobs

C. 125 jobs

D. 0.008 jobs

Correct Answer: A. 20 jobs

Question 5

Available production time is 360 minutes and demand is 90 units. The takt time is:

A. 0.25 minute per unit

B. 4 minutes per unit

C. 40 minutes per unit

D. 450 minutes per unit

Correct Answer: B. 4 minutes per unit

Question 6

The main purpose of line balancing is to:

A. Increase task variation

B. Assign tasks to stations while meeting pace and precedence constraints

C. Eliminate all maintenance

D. Replace quality inspection

Correct Answer: B. Assign tasks to stations while meeting pace and precedence constraints

Question 7

Finite-capacity scheduling differs from infinite-capacity scheduling because it:

A. Ignores working calendars

B. Assumes unlimited machines

C. Recognises actual resource and skill limits

D. Uses no priority rules

Correct Answer: C. Recognises actual resource and skill limits

Question 8

If a job has a due date of 12 hours, current time is 8 hours and remaining processing time is 5 hours, its critical ratio is:

A. 0.80

B. 1.25

C. 4.00

D. 5.00

Correct Answer: A. 0.80

Question 9

Which is an appropriate workforce-planning control?

A. Ignoring competence differences

B. Using a skills matrix and checking legal working limits

C. Scheduling all employees at maximum overtime

D. Removing rest periods

Correct Answer: B. Using a skills matrix and checking legal working limits

Question 10

A Gantt chart is most directly useful for showing:

A. The organisation’s annual profit

B. Activities, resources and timing across a schedule

C. Employee personality types

D. Supplier financial statements

Correct Answer: B. Activities, resources and timing across a schedule