PMGT1865 Chap.3 Sequencing Activities into a Precedence Network
Sequencing Activities into a Precedence Network
Fixing the dependencies and then building the schedule are the two steps that convert a bare task list into a timeline somebody can work to. Dependency has one specific meaning here: working out which pieces of work wait on the beginning or the completion of others. Week 2 sets out a five step process from a work list to a dated schedule, and this chapter covers the first four of those steps.
Notice that resources are estimated before durations, because how long an activity takes depends on how many people are on it. The chapter also fixes the node used throughout the unit, a six cell box in which only the identifier, the description and the duration are given and everything else is a result, and it sets the five placement rules that produce a diagram somebody else can check.
Two node types matter more than the rest: merge activities, which have several arrows coming in, and burst activities, which have several going out. Every choice either pass makes happens at one of those two.
What this chapter covers
- 01
Dependency as a claim about what physically or contractually stops work starting sooner
- 02
The four factors that cap how fast a project can go, including resource availability at a specific point
- 03
The five step process, and why activity resources are estimated before activity durations
- 04
Step one: selecting a planning level and assigning unique identifiers from the breakdown
- 05
The six cell node, which cells are given and which are computed
- 06
Why networks run on elapsed working periods counted from zero rather than on calendar dates
- 07
The five placement rules, including the no dead ends rule and its single exception
- 08
Merge and burst activities, and why each pass makes its only decision at one of them
- 09
Duration based against effort based estimating, and the division that connects them
- 10
Who should estimate, why estimates exclude contingency, and progressive elaboration
Drawing and validating a precedence network
- +1Create seven nodes, one per row, each carrying its identifier, description and duration. Leave the four date cells and the float cell empty, because those are results rather than inputs.
- +1Place A on the left. It is the only row with an empty predecessor cell, so it is the single start node.
- +1Place B, C and E to the right of A with one arrow from A into each. A is therefore a burst activity with three successors.
- +1Place D to the right of B and C with arrows from both. D is a merge activity, and since C runs eight days against B at six, C is the predecessor that will govern its start.
- +1Place F to the right of D and E with arrows from both, then G to the right of F. F is the second merge activity and G is the only legitimate dead end.
- +1Run the two sweeps. Every node except G has an outgoing arrow, every node except A has an incoming one, and no arrow runs right to left. The diagram is ready for a forward pass.
Key terms
- Predecessor
- An activity that must reach a defined point, usually its finish, before another activity may begin. The relationship is a claim about the work, not a statement of the order somebody intends to work in.
- Merge activity
- A node with more than one arrow arriving. It cannot start until all its predecessors are satisfied, so the forward pass takes the highest early finish among them.
- Burst activity
- A node with more than one arrow leaving. It must be finished in time for the earliest of its successors, so the backward pass takes the lowest late start among them.
- Effort based estimating
- Starting from the number of person hours, days or weeks of work and dividing by the resources assigned to obtain a duration. It makes explicit the lever that duration based estimating hides.
- Progressive elaboration
- Revising estimates regularly as the project proceeds and more becomes known. It is why an estimate is treated as a forecast with a date on it rather than as a promise made once.
Sequencing Activities into a Precedence Network FAQ
How do I tell a real dependency from a preference?
Ask what physically or contractually stops the second activity starting sooner. If two activities could genuinely run at the same time given the people available, they are parallel, and drawing them in series invents a longer project and hides the real critical path.
The most common source of false dependencies is the resource plan: two activities linked because the same person does both is a resource constraint, and it belongs in the levelling chapter rather than in the logic.
Why does the process estimate resources before durations?
Because duration is effort divided by the resources assigned, so a duration is not a property of an activity on its own. Two carpenters and a crane produce a different number from four carpenters and no crane. Recording which you assumed is what lets you answer the obvious follow up question about whether adding people would help, and a plan that never wrote the assumption down cannot answer it.
What does the no dead ends rule actually catch?
An activity with no successor is claiming that nothing in the project waits on it, and that is true of exactly one node. Everywhere else it means a link was forgotten or the activity does not belong in the project. The error is worth sweeping for before the forward pass, because a dangling node produces no warning: it simply lets the project finish earlier on paper than it can in practice.
Should estimates include a safety margin?
No. Estimates should reflect schedule risk, which means realistic rather than hopeful, and they should not include contingency time. Contingency is held once, visibly, at project level where a manager can see it and spend it deliberately. Padding buried inside every activity is invisible, unmanageable and spent by default, because work expands to fill whatever time it is given.
Assessment move
Practise the mechanical parts until they are automatic, because they are the parts a diagram is checked on. Take any dependency table and draw the network three times: once to find the shape, once tidied with chains horizontal, and once from memory to prove you can. Every time, circle the merge and burst nodes as the last step, because doing so converts both later passes from a sequence of judgements into bookkeeping.
Then check the diagram the way a marker will: every arrow left to right, exactly one node with no predecessor and one with no successor, table rows and diagram arrows agreeing activity by activity, durations transcribed correctly, and a legend showing which node cell is which. Four of those five are transcription, and transcription errors are the expensive kind here because everything downstream inherits them.
Keep the convention that early start begins at zero, write it in your legend, and never mix it with a one based convention inside the same assignment.
Working through Sequencing Activities into a Precedence Network in PMGT1865? Sia is AskSia’s AI Project Management tutor — ask any PMGT1865 Sequencing Activities into a Precedence Network question and get a clear, step-by-step explanation grounded in how PMGT1865 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
