Sequencing and logic: dependencies, constraints and lags

Once the activities and their durations are known, they must be put in order. Logic, the network of relationships between activities, is what turns a list into a schedule. It records which work must finish before other work can start, which work can overlap, and by how much. Good logic reflects how the job will really be built. Bad logic, whether missing, wrong or forced with date constraints, makes the critical path meaningless and every forecast unreliable. This lesson covers the kinds of dependency, the four relationship types, lags and leads, constraints, and the diagramming conventions used in construction.

Why activities depend on each other

Dependencies in construction come from several sources, and a good scheduler knows which kind each link represents.

• Physical or mandatory dependencies come from the nature of the work: footings before foundation walls, structure before roofing, rough-in before drywall. They cannot be changed without changing the design or method.
• Resource dependencies arise because the same crew, crane or set of forms is used in sequence: the same forming crew moves from area A to area B, so B cannot start until A is done, even though B does not physically need A.
• Preferential or discretionary dependencies reflect the team's chosen method: finishing level 2 before level 3 so the drywall crew can work top-down, for instance. They can be changed if the plan changes.
• External dependencies involve parties outside the project team: a utility connection, a permit, an inspection, owner-supplied equipment.
• Safety and regulatory dependencies: work below must not proceed while overhead work continues, excavations must be inspected before entry, fire protection must be operational before occupancy.

Separating these matters because discretionary and resource logic can be changed to recover time, while physical logic generally cannot. When a schedule is under pressure, the first place to look for time is in the discretionary links.

The precedence diagramming method

Modern construction schedules use the precedence diagramming method, in which each activity is a box, or node, and arrows between boxes show relationships. It is sometimes called activity-on-node. An older method, activity-on-arrow or the arrow diagramming method, drew each activity as an arrow between event circles and needed dummy activities to show some logic. Activity-on-arrow is now mostly of historical interest, though its vocabulary, such as events and nodes, survives.

In the precedence method, the relationship between two activities is described by its type and an optional lag. The predecessor is the activity at the tail of the arrow, and the successor is at the head.

The four relationship types

Finish-to-start is the default and should make up most of a schedule's links. The Defense Contract Management Agency's 14-point schedule assessment, widely used as a quality benchmark in the United States and elsewhere, expects at least 90 % of relationships to be finish-to-start. Start-to-start and finish-to-finish are useful for overlapping work, especially on repetitive or linear tasks. Start-to-finish is rare and often confusing; many schedulers avoid it entirely.

Lags and leads

A lag is a delay added to a relationship. A finish-to-start relationship with a 7-day lag means the successor can start 7 days after the predecessor finishes, for example to allow concrete to cure before stripping forms. A start-to-start relationship with a 2-day lag means the successor can start 2 days after the predecessor starts, so that the first crew has opened up enough work for the second to follow.

A lead is a negative lag: the successor can start before the predecessor finishes. A finish-to-start relationship with a 3-day lead means the successor can start 3 days before the predecessor finishes. Leads are generally discouraged in professional scheduling practice and many specifications forbid them, because the same overlap can be shown more clearly with a start-to-start relationship and a positive lag, or by splitting the activity. The DCMA assessment expects no leads at all and lags on no more than about 5 % of relationships.

Lags are also abused. A long lag used to represent work, such as a 20-day lag standing in for fabrication, hides an activity that should be in the schedule, cannot be progressed, and makes the critical path hard to follow. Lags should represent waiting time, such as curing or drying, not work.

Constraints

A constraint is a date imposed on an activity instead of, or in addition to, logic. Scheduling software offers several kinds.

• Start no earlier than, and finish no earlier than: the activity cannot start or finish before a date, for example because a permit or access date is fixed. These are soft constraints: they still let logic push the activity later.
• Start no later than, and finish no later than: the activity must start or finish by a date, for example an interim milestone in the contract. If logic pushes the activity past the date, negative float appears, warning that the date will be missed.
• Mandatory start and mandatory finish, sometimes called must start on or must finish on: the activity is fixed to a date regardless of logic. These are hard constraints that break the logic of the network, and they can make a late schedule look on time.

Constraints have legitimate uses: contract milestones, access dates, and the delivery date of owner-supplied equipment. But they should be few. The DCMA assessment expects hard constraints on no more than about 5 % of activities, and many specifications prohibit mandatory constraints entirely in schedule submittals. Every constraint should be justified in the schedule narrative.

Open ends and dangling activities

In a well-built network every activity except the project start has at least one predecessor, and every activity except the project finish has at least one successor. An activity with no successor is an open end: if it slips, nothing downstream moves, so the schedule cannot show the effect of the delay. The DCMA assessment calls missing logic a key defect and expects it on no more than about 5 % of incomplete activities.

A subtler problem is a dangling activity: one that has a predecessor driving its start but no successor tied to its finish, or the reverse. Suppose drywall taping has a start-to-start link from hanging but nothing linked to its finish. If taping takes twice as long as planned, the painting that follows it does not move. Every activity should have logic driving its start and logic depending on its finish.

Logic for typical building work

Every project is different, but much building logic follows familiar patterns that experienced superintendents carry in their heads. A simplified sequence for a low-rise building shows how physical, resource and discretionary logic combine.

• Site: mobilization, clearing and layout, excavation, underground services.
• Substructure: footings, foundation walls, waterproofing and drainage, backfill, underslab services, slab on grade.
• Structure: columns and beams or walls, floor and roof decks, concrete toppings.
• Dry-in: roofing, exterior walls and windows, temporary closures, so that interior work is protected from weather.
• Rough-in: mechanical, electrical, plumbing and fire protection rough-in, often overlapping with interior framing.
• Close-in: inspections, insulation, drywall hanging, taping and finishing.
• Finishes: ceilings, painting, flooring, millwork, doors and hardware, specialties, with trim-out of MEP systems.
• Completion: testing and commissioning, cleaning, deficiency correction, occupancy permit, substantial completion and closeout.

Dry-in deserves special attention. Many interior finishes cannot proceed until the building is weathertight and heated, so roofing and windows often lie on or near the critical path, and late window deliveries are a classic cause of delay.

Reviewing logic

Before the schedule is calculated, the logic should be reviewed with the people who will build the work. A practical review walks through the network area by area and asks the superintendent and trades: Can this really start then? What has to be in place first? Who else is in the space? Where are the inspections? What happens if this is late? Many errors, such as a missing inspection hold point, a trade that needs a crane already committed elsewhere, or a sequence that cannot work around a tower crane base, are caught only this way.