Schedule Development and Critical Path Analysis

Introduction

A project schedule is a time-based model of the work required to produce project deliverables. It translates scope into activities, establishes logical relationships, estimates durations, assigns resources, and identifies dates against which progress can be controlled. A schedule is more than a list of tasks: it is a defensible explanation of how the project is expected to move from authorization to completion.

In software projects, scheduling is especially challenging because requirements may evolve, productivity varies, technical uncertainty is significant, and many activities depend on scarce specialists. A useful schedule therefore combines analytical techniques—such as network analysis and the Critical Path Method (CPM)—with professional judgement, stakeholder agreement, and controlled change.

Learning Objectives

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

• distinguish milestones, activities, dependencies, and work packages;
• construct milestone charts, Gantt charts, and precedence networks;
• calculate early dates, late dates, total float, and the critical path;
• apply PERT estimates when activity durations are uncertain;
• interpret resource and schedule constraints;
• establish and maintain a credible schedule baseline; and
• communicate schedule risks and changes professionally.

1. From Scope to a Schedule Model

Schedule development begins with the scope baseline. The work breakdown structure (WBS) decomposes deliverables into work packages. Activities are then defined as the manageable units of work needed to produce those work packages. A work package is a scope element; an activity is an action that consumes time and usually resources.

Important terms include:

• Activity: A scheduled piece of work with a defined start and finish.
• Milestone: A significant event with zero planned duration, such as approval or release.
• Predecessor: An activity that logically comes before another activity.
• Successor: An activity that follows another activity.
• Dependency: A logical relationship between activities.
• Duration: The amount of working time required by an activity under stated conditions.
• Calendar: The working-time rules, including working days, holidays, shifts, and time zones.
• Schedule baseline: The approved, time-phased version of the schedule used for comparison with actual performance.

Activity definition should be sufficiently detailed to support estimation and control, but not so detailed that the schedule becomes expensive to maintain. A software activity such as “build system” is too broad for effective control. Activities such as “implement authentication,” “write integration tests,” and “obtain security review” are more observable and estimable.

Dependency Types

The most common logical relationship is finish-to-start (FS): a successor cannot start until its predecessor finishes. Other relationships are:

• Start-to-start (SS): the successor cannot start until the predecessor starts;
• Finish-to-finish (FF): the successor cannot finish until the predecessor finishes; and
• Start-to-finish (SF): the successor cannot finish until the predecessor starts.

A relationship may include a lead or lag. A two-day lag on a finish-to-start relationship means the successor begins two working days after the predecessor finishes. Leads permit overlap, but they should be used cautiously because excessive leads can conceal real coordination requirements.

Dependencies may be mandatory, based on the nature of the work; discretionary, based on a preferred method; external, involving parties outside the project; or internal, controlled by the project team. Discretionary dependencies should be recorded because changing them may provide a legitimate schedule-compression option.

2. Schedule Representations

Milestone Chart

A milestone chart displays major events against calendar time. Typical milestones include requirements approval, architecture approval, completion of system testing, user acceptance, and production release. It is useful for governance audiences because it communicates outcome dates without overwhelming them with task-level detail.

A milestone is not a substitute for a schedule. It does not show the work needed to reach the event or the logical causes of delay.

Gantt Chart

A Gantt chart displays activities as horizontal bars on a time scale. It can show planned dates, actual progress, dependencies, milestones, and a baseline. Its main advantages are accessibility and ease of communication. Its limitations are that dense dependency networks can become difficult to read and that a bar does not, by itself, explain why an activity has its duration or priority.

A professional Gantt chart should identify the calendar, status date, activity owners, baseline dates, and the meaning of colours or symbols. Percent complete must be defined carefully; “50% complete” may mean half the effort, half the deliverables, or merely that the activity has begun.

Network Diagram

A precedence network emphasizes logical relationships rather than calendar appearance. In an activity-on-node network, each node represents an activity and arrows represent dependencies. Network analysis supports calculation of the earliest and latest permissible dates and reveals the critical path.

A valid network should have a clear start and finish, no unexplained cycles, and no activities that are disconnected from the project outcome. Relationships should express logic, not merely a desire to make bars appear sequential.

3. Critical Path Method

CPM is a deterministic technique. It assumes that each activity has one estimated duration and uses network logic to calculate schedule dates. The critical path is the longest path through the network in terms of total duration. Activities on that path have zero total float under the stated model, although the critical path may change when durations, logic, calendars, or resources change.

For a simple finish-to-start network, define:

• ES = early start;
• EF = early finish;
• LS = late start;
• LF = late finish; and
• d = activity duration, measured in consistent working-time units.

The forward-pass equations are:

For an activity with several predecessors:

The backward-pass equations are:

For an activity with several successors:

The total float is:

These equations assume a single project-start reference, a finish-to-start network, no resource constraints, and durations expressed in the same calendar units. If the schedule uses inclusive date conventions, midnight boundaries, or non-working calendars, the software’s date arithmetic must be checked; simply adding calendar numbers can produce off-by-one errors.

Worked CPM Example

Consider the following illustrative network. Durations are working days, and the project starts at time 0.

Forward pass. A has ES 0 and EF 3. B has ES 3 and EF 7. C has ES 3 and EF 5. D has ES 7 and EF 13. E has ES 5 and EF 8. F must wait for both D and E, so its ES is max(13, 8) = 13 and its EF is 17. G has ES 17 and EF 19. The planned project duration is therefore 19 working days.

Backward pass. Set G’s LF to 19, giving LS 17. F has LF 17 and LS 13. D has LF 13 and LS 7. E has LF 13 and LS 10. B has LF 7 and LS 3. C has LF 10 and LS 8. A has LF equal to the smaller of B’s LS 3 and C’s LS 8, so A has LF 3 and LS 0.

The critical path is A–B–D–F–G, with duration 19 working days. C and E each have five working days of total float. Float is not automatically “free time”: consuming it may reduce flexibility for later activities or create resource conflicts.

Types of Float

Total float is the amount an activity can slip without delaying the calculated project finish, assuming other conditions remain unchanged. Free float is the amount an activity can slip without delaying the early start of any immediate successor. For a finish-to-start relationship, a common expression is:

where FF is free float. Free float cannot exceed total float in a conventional CPM network. A schedule may also show interfering float, independent float, or negative float, depending on the scheduling method and constraints.

Negative float occurs when a required completion date is earlier than the date implied by the network. It is not spare capacity; it is evidence that the schedule currently cannot meet the imposed constraint without changing scope, logic, resources, calendars, or duration assumptions.

4. PERT and Uncertainty

CPM is useful when activity durations are sufficiently stable. For uncertain work, the Program Evaluation and Review Technique (PERT) uses three estimates:

• optimistic time, a: favourable but realistic conditions;
• most-likely time, m: the normal expected condition; and
• pessimistic time, b: adverse but plausible conditions.

The traditional PERT expected duration is:

The associated variance is:

Here te is expected duration, σ² is duration variance, and all three estimates must use the same time unit. The formula assumes a beta-shaped approximation in which the most-likely estimate receives four times the weight of each extreme. It is a planning approximation, not a guarantee of actual performance.

For example, if a testing activity has a = 2 days, m = 5 days, and b = 8 days:

te = (2 + 4×5 + 8) ÷ 6 = 30 ÷ 6 = 5 days

and

σ² = ((8 − 2) ÷ 6)² = 1 day².

PERT can improve discussion of uncertainty, but it does not remove uncertainty. Activity durations may be correlated: a difficult technology can delay several activities simultaneously. In such cases, adding individual variances as though activities were independent can understate risk. Simulation, such as Monte Carlo analysis, is more appropriate when the organization has reliable probability distributions, relationships, and historical data.

5. Resource and Calendar Feasibility

A technically valid network can still be impossible to execute. CPM’s basic calculations generally assume unlimited resources. If two activities require the same specialist at the same time, the schedule may require resource levelling or resource smoothing.

• Resource levelling changes activity dates, and possibly the project finish, to resolve resource over-allocation.
• Resource smoothing uses available float to reduce variation without normally changing the critical completion date.

The schedule should identify resource availability, public holidays, part-time assignments, time zones, planned leave, procurement lead times, and external approval periods. A duration is not the same as effort. For example, 24 person-hours may require three days if one person works eight hours per day, but it may not require one day with three people if the work is sequential or communication overhead is high.

Constraints should be classified. A contractual deadline may be hard; a preferred release date may be soft. Excessive use of “must start on” or “must finish on” constraints can hide the actual network and weaken analysis. A defensible schedule records the reason, owner, and evidence for each significant constraint.

6. Schedule Compression

When a date is unacceptable, the team may consider schedule compression. Crashing adds resources, overtime, specialist support, or expenditure to shorten an activity. It is viable only when additional input can reduce duration and when the cost and quality effects are acceptable. Fast tracking overlaps activities that were originally planned sequentially. It increases coordination and rework risk and should be used only when the work can safely overlap.

Compression should focus first on the current critical path. Shortening a non-critical activity may produce no reduction in project duration. After every change, the network must be recalculated because a different path may become critical. Compression decisions should document cost, quality, risk, safety, and stakeholder impacts—not merely the new date.

7. Establishing a Defensible Schedule Baseline

A schedule baseline is established only after sufficient review and approval. A practical baseline package includes:

1. the approved scope and WBS;
2. activity names, codes, owners, durations, calendars, and assumptions;
3. logical relationships and external dependencies;
4. planned start and finish dates;
5. milestones and required completion dates;
6. resource and cost assumptions where relevant;
7. identified risks, constraints, and schedule contingency; and
8. approval records and the baseline version.

The baseline should be realistic rather than artificially optimistic. It should be traceable to estimates, historical data, expert judgement, or documented assumptions. Stakeholders should understand whether dates represent working days or calendar days and whether contingency is included within activities, held as a reserve, or represented separately.

Once approved, actual progress is recorded against the baseline. A baseline is not frozen reality; it is a controlled reference. Approved changes may result in a revised baseline, but silently overwriting planned dates destroys the evidence needed to learn from performance. Baseline changes should have a change request, impact assessment, authorization, effective date, and version history.

Schedule performance can be reviewed using simple indicators. Schedule variance in earned value management is:

where EV is earned value and PV is planned value, both expressed in monetary or equivalent value units. A negative SV indicates less value has been earned than planned, but SV is not a direct measure of days late. The schedule performance index is:

An SPI below 1.00 generally indicates that progress is behind the planned value profile. These measures require a credible measurement system and should not replace examination of critical activities and remaining work.

8. Applied Case: Online Permit Service

The following is an illustrative case, not a report of a real organization.

Background

A municipal technology team must deliver an online permit service before the beginning of a new regulatory year. The service includes requirements confirmation, user-interface design, application development, security testing, user acceptance, training, and deployment.

Problem

The sponsor requests a firm launch date, but security staff are shared across several projects. Earlier plans listed broad phases without showing dependencies. The sponsor cannot determine whether a two-week delay in interface design will affect launch.

Analysis

The project manager decomposes the work into activities and identifies mandatory security approval before production deployment. A network analysis shows that requirements, architecture, development, integration testing, security testing, acceptance, and deployment form the current critical path. Training has some float, but only if training materials are completed before the planned sessions.

The team uses three-point estimates for security testing because the technology is unfamiliar. With estimates of 4, 6, and 12 working days, the PERT expected duration is (4 + 4×6 + 12) ÷ 6 = 40 ÷ 6 ≈ 6.67 days. The manager rounds only for calendar scheduling after retaining the unrounded estimate for analysis. A risk register records the possibility that security findings will create rework.

Approach and Solution

The team develops a network schedule, assigns the security specialist’s actual availability calendar, and produces a milestone chart for the sponsor. It establishes a baseline after the security lead and product owner approve the assumptions. A two-day management reserve is identified outside the activity estimates for recognized uncertainty, while unknown changes remain subject to change control.

When interface design slips by three days, the team does not immediately move the launch date. It checks float, resource conflicts, and the effect on integration testing. Because the design activity had two days of float, the first two days are absorbed; the remaining day affects development and is escalated. The team evaluates limited fast tracking between training preparation and late-stage testing, but rejects overlapping activities that would create unacceptable compliance risk.

Outcome

The launch date remains achievable after an approved sequence adjustment and additional review coordination. The baseline remains intact, while the approved change is recorded with its impact. The sponsor receives an honest forecast rather than an unexplained revised bar chart.

Learning Points

The case demonstrates that a schedule is a model for decision-making. Clear logic, explicit calendars, realistic uncertainty, and controlled baselines allow managers to distinguish a harmless float consumption from a threat to the launch date.

9. Limitations, Governance, and Contemporary Practice

Schedules are forecasts, not promises of certainty. Common limitations include inaccurate estimates, missing dependencies, optimism bias, changing scope, unrecorded work, resource contention, and false precision in software dates. Agile teams may use release plans, iteration plans, product backlogs, and cumulative-flow information rather than a fully detailed long-range task network. Nevertheless, the same principles remain relevant: define outcomes, expose dependencies, forecast capacity, identify risks, and establish a transparent reference point.

Scheduling tools increasingly provide automatic critical-path calculation, resource analysis, scenario comparison, and probabilistic forecasting. Artificial intelligence may assist with estimating durations, detecting missing relationships, or identifying patterns in historical schedules. Such outputs require human review because training data may encode organizational bias, omit exceptional events, or treat correlation as causation. Project information may also contain commercially sensitive or personal data; access, retention, auditability, and vendor governance must be addressed.

Ethically, schedule pressure must never justify unsafe working practices, concealment of defects, falsification of progress, or deliberate underestimation. A professional schedule communicates uncertainty and adverse information early. Governance bodies should challenge unrealistic deadlines, require traceable assumptions, and ensure that quality, security, accessibility, and regulatory activities are not removed merely to preserve an attractive date.

Best-Practice Checklist

• Derive activities from an approved WBS and deliverables.
• Use logical relationships that represent real work conditions.
• Document assumptions, calendars, constraints, and external dependencies.
• Estimate duration separately from effort and resource availability.
• Calculate and review the critical path after significant changes.
• Use PERT or simulation when uncertainty justifies it; do not imply false precision.
• Validate resource feasibility with the people who perform the work.
• Present milestones to executives and detailed networks to delivery teams.
• Baseline only after review and approval.
• Preserve the baseline and control changes through versioned governance.
• Report forecast dates and risks honestly, including negative float.

Chapter Summary

Schedule development converts project scope into a time-phased model of activities, dependencies, resources, milestones, and constraints. Milestone charts communicate major events, Gantt charts communicate timing and progress, and network diagrams expose logical structure. CPM uses forward and backward passes to calculate early and late dates, float, and the critical path. PERT incorporates optimistic, most-likely, and pessimistic estimates when duration uncertainty is material. Resource calendars, dependencies, and constraints determine whether a mathematically valid schedule is practically feasible. A defensible baseline is realistic, traceable, approved, versioned, and protected from unauthorized overwriting. Professional scheduling combines quantitative analysis with judgement, ethical communication, and controlled change.

Key Terms

Activity: A defined unit of scheduled work.

Baseline: The approved schedule used for performance comparison.

Calendar: Rules defining working and non-working time.

Critical path: The longest logical path determining calculated project duration.

Dependency: A logical relationship between activities.

Duration: Working time required to complete an activity.

Early finish: Earliest calculated completion time.

Early start: Earliest calculated start time.

Float: Permitted schedule flexibility under stated assumptions.

Gantt chart: A time-scaled bar chart of activities.

Lag: Deliberate waiting time in a dependency.

Milestone: A zero-duration significant event.

Network diagram: A graphical representation of activities and relationships.

PERT: A three-point estimating technique for uncertain durations.

Predecessor: An activity that logically comes before another.

Resource levelling: Adjusting dates to resolve resource over-allocation.

Schedule compression: Reducing duration through crashing or fast tracking.

Total float: Delay permitted without delaying the calculated project finish.

Work breakdown structure: Hierarchical decomposition of project scope.

Review Questions

Short Answer Questions

1. What is the difference between an activity and a milestone?
2. Name and define the four standard dependency types.
3. What does total float measure?
4. State the PERT expected-duration formula.
5. Why should a schedule baseline not be silently overwritten?

Long Answer Questions

1. Explain how a WBS is transformed into a network schedule and then represented in a Gantt chart.
2. Describe the forward-pass and backward-pass procedures in CPM, including their assumptions.
3. Compare CPM and PERT and explain when each is appropriate.
4. Explain how resource calendars and resource levelling can change a logically valid schedule.
5. Discuss the contents and governance requirements of a defensible schedule baseline.

Analytical / Application Questions

1. Using the activities A–G in the worked example, explain why delaying activity E by four working days does not necessarily delay the project, but delaying activity D by two working days does.
2. A project has optimistic, most-likely, and pessimistic estimates of 3, 7, and 15 days for a security review. Calculate its PERT expected duration and variance, and interpret the result.
3. A sponsor imposes a completion date earlier than the calculated finish, producing negative float. Recommend a structured sequence of actions before promising the date.

Multiple Choice Questions with Answers

Question 1

Which schedule representation is primarily designed to communicate major events and approval points to governance stakeholders?

A. Milestone chart

B. Resource histogram

C. Risk register

D. Requirements traceability matrix

Correct Answer: A. Milestone chart

Question 2

In a finish-to-start network, an activity with ES = 8 and duration d = 5 has which early finish?

A. 3

B. 8

C. 13

D. 40

Correct Answer: C. 13

Question 3

For an activity with ES = 6 and LS = 9, what is its total float under conventional CPM?

A. 3 time units

B. 6 time units

C. 9 time units

D. 15 time units

Correct Answer: A. 3 time units

Question 4

Which statement best defines the critical path?

A. The path containing the most activities

B. The longest-duration logical path determining calculated project duration

C. The path with the highest financial cost

D. The path selected by the project sponsor

Correct Answer: B. The longest-duration logical path determining calculated project duration

Question 5

In PERT, the most-likely estimate receives what weighting in the traditional expected-duration formula?

A. One

B. Two

C. Four

D. Six

Correct Answer: C. Four

Question 6

For PERT estimates a = 2, m = 5, and b = 8 days, what is the expected duration?

A. 2 days

B. 4 days

C. 5 days

D. 8 days

Correct Answer: C. 5 days

Question 7

Which technique overlaps activities that were previously planned sequentially?

A. Crashing

B. Fast tracking

C. Resource smoothing

D. Backward pass

Correct Answer: B. Fast tracking

Question 8

What does negative float usually indicate?

A. The project has excess schedule capacity

B. The imposed completion date is earlier than the network-based date

C. All activities are ahead of plan

D. The schedule has no constraints

Correct Answer: B. The imposed completion date is earlier than the network-based date

Question 9

Which item is most important for preserving the integrity of a schedule baseline?

A. Overwriting old planned dates whenever actual dates change

B. Removing activities with poor performance

C. Maintaining approved versions and documenting authorized changes

D. Using only calendar days in every project

Correct Answer: C. Maintaining approved versions and documenting authorized changes

Question 10

Which statement about CPM is correct?

A. It normally assumes one deterministic duration for each activity

B. It requires three probability estimates for every activity

C. It ignores logical relationships

D. It guarantees that the project will finish on the calculated date

Correct Answer: A. It normally assumes one deterministic duration for each activity