(Article published first in “The Critical Path”, PMI Sydney, February 2015)

As argued in a previous article^[1], common definitions of œdeliverable in the WBS and schedule network may be maintained by having œStart and œFinish milestones in the schedule network for each WBS summary component. This article introduces a œSchedule Network 100 percent Rule and the œAdd and Prune Dependencies Algorithm by which we can automate this coherence between the WBS and schedule network at project scale. The Schedule Network 100 percent Rule is applied to a WBS to derive an initial schedule network. Required dependencies are then added to the initial schedule network and redundant dependencies are pruned, using the newly introduced œAdd and prune dependencies algorithm. The result is a sequenced schedule network that is coherent with the WBS. A following article will introduce open source software that implements this algorithm and supports the progressive elaboration of a project plan under rolling wave planning.

The 100 percent rule

The Schedule Network 100 percent Rule allows us to construct an initial, fully connected, schedule network based solely on the WBS. The well-known WBS 100 percent rule states^[2]: œThe next level decomposition of a WBS element (child level) must represent 100 percent of the work applicable to the next higher (parent) element. Both the WBS and the schedule network are œmaps of the same territory and use different visual languages to represent the deliverables of the project. This article introduces^[3] the Schedule Network 100 percent Rule which may be stated as: œA summary activity’s start milestone must precede, and its finish milestone must succeed, 100 percent of the work applicable to the summary activity.

Let’s apply the Schedule Network 100 percent Rule to the sample WBS in Diagram 1.

We create a schedule network with œStart Project and œFinish Project milestones before, and after, 100 percent of the project’s work. We place Start and Finish milestones before, and after, 100 percent of A’s work, and before and after A.2’s work. The result is a fully connected, but unsequenced, schedule network (Diagram 2.) that is coherent with the WBS

 

Schedule Network 100% Rule animation

Click on the play button for a step-by-step animation of and example application of the Schedule Network 100% Rule.

Add and Prune Dependencies

The new^[4] Add and Prune Dependencies Algorithm allows us to automatically construct a fully sequenced and simplified schedule network coherent with the WBS.

Often, activities and planning packages are sequenced and then open ends are joined to form a completed schedule network, however, when we do this, coherence with the WBS is not preserved. Instead, we start with a fully connected, but unsequenced schedule network (Diagram 2. above) and use the œAdd and Prune Dependencies algorithm to sequence the activities and simplify the plan while ensuring that the Schedule Network 100 percent Rule is maintained.

The Add and Prune Dependencies algorithm has two phases:

1. Add a dependency, and then
2. Simplify by pruning any other dependencies that are now redundant and are no longer required by the Schedule Network 100 percent Rule.

In our example the schedule network needs to be sequenced with the following dependencies (Diagram 3.).

Adding the first dependency (A.1 to A.2 Start) and applying the Add and Prune Dependencies Algorithm gives us the Schedule network shown in Diagram 4. It shows the added dependency as well as the dependencies that have become redundant and marked for pruning. The two dependencies may be pruned because they are no longer necessary for maintaining the Schedule Network’s 100 percent Rule. The dependency from œA Start to œA.2 Start may be pruned because the newly added dependency ensures that œA Start precedes œA.2 Start.  Similarly, the dependency from œA.1 to œA Finish may be pruned because the newly added dependency ensures that œA Finish succeeds œA.1.

After adding the remaining two dependencies (from A.2.1 to A.2.2 and from A.2.1 to B) and automatically pruning three redundant dependencies, we get a fully sequenced schedule network (Diagram 5.) that continues to satisfy the Schedule Network 100% Rule and is therefore coherent with the WBS. Critical path and other scheduling algorithms may now be run on the completed network.

 

Add and Prune Dependencies Algorithm animation

Click on the play button for a step-by-step animation of the Add and Prune Dependencies Algorithm.

This article has introduced a new Schedule Network 100% Rule and the Add and Prune Dependencies Algorithm and show how they, together, allow us to derive a schedule network that incorporates required dependencies and is both 1) as simple as possible, and 2) fully consistent with the WBS. This ensures that the WBS and schedule network share a common vocabulary of œdeliverables and automatically ensures that the schedule network avoids open ends. A following article will introduce open source add-in to MS Project that implements this rule and algorithm.

^[1] Pratten, D. (2014). Having a common definition of œdeliverable in both WBS and the schedule network The Critical Path, 3(6), 16-17.

[2] Haugan, Gregory T. (2002). Effective Work Breakdown Structures. Vienna, VA Management Concepts.

[3] No prior descriptions of the Add and Prune Dependencies Algorithm were found in the existing Project Management literature.

[4] No prior descriptions of the Schedule Network 100 percent Rule were found in the existing Project Management literature.

(Article published first in “The Critical Path”, PMI Sydney, December 2014)

Having a common definition of œdeliverable in both WBS and the schedule network assists project managers to build a common understanding of the project scope and progress amongst stakeholders. This article examines possible causes for the divergence of the schedule network’s definitions and those of the WBS and argues that it is best practice to follow the example of Figure 6-21 in the PMBOK(r) Guide 5th Ed. and include a œStart and œFinish milestone in the schedule network for each WBS component. At project scale, implementing this practice requires automation help and a following article will explain a new project management algorithm œAdd and Prune Dependencies which enables this.

The project model has, at its heart, the dual structures of the tree-like Work Breakdown Structure (WBS) and the schedule network. Both the WBS and the project schedule are œmaps of the same territory and use different visual languages to represent the deliverables of the project. In the WBS the top-level component represents 100% of the deliverables or products of the project, and each level below decomposes the same 100% into smaller components that represent part of the deliverables. In the schedule network 100% of the deliverables of the project are represented between the œStart Project and œFinish Project milestones. Likewise, smaller deliverables may be represented by pairs of milestones that mark the start and completion of these deliverables.

Outside of domains that rigorously apply Earned Value Management (EVM), it is common for the deliverables defined in the WBS to not be precisely traceable to milestones in the schedule network. Assuming that dependencies are not placed on, so called, œSummary tasks, the planner will commonly create œDeliverable completed milestones in the schedule network to represent the availability of interim deliverables and the point where work that depends on those interim deliverables can begin. However, manually created œDeliverable completed milestones are likely (and with updating over time, inevitable) to not exactly correspond to the completion of any WBS component. There are two main reasons for this. Firstly, the selection of predecessor relationships for the œDeliverable completed milestone is unlikely to correspond exactly with the set of predecessors required to match the completion of a WBS component, and secondly, the choice of deliverables highlighted by œDeliverable completed milestones may be dominated by tactical considerations at a detail planning level rather than the set of deliverables already defined in the WBS. As the project progresses, this can result in a widening mismatch between the WBS œmap and the schedule network œmap.

So how can a WBS and schedule network be provably based on the same set of deliverables?

Figure 1 shows the simplest case of a two level WBS with two planning packages at the lowest level and its corresponding schedule network prior to sequencing. Notice that the œHouse deliverable is represented in the schedule network by two milestones œStart House and œFinish House. In general this is true. The implication is that if we want to have common definitions of œdeliverable in both the WBS and schedule network, each component in the WBS will be mirrored by œStart and œFinish milestones in the schedule network.

The thesis of this article is that definitions of deliverables will inevitably diverge if the schedule does not maintain œStart and œFinish milestones matching each WBS component, and that their presence in the schedule will create the alignment of definitions that we are looking for. Interestingly, while PMI’s best practice guidance is largely silent on this question, Figure 6-21 on p183 the PMBOK(r) Guide 5th Ed. shows an example of exactly this with œStart and œFinish milestones in the schedule network that correspond to WBS components.

Outside of domains that rigorously apply EVM this is not (yet) common practice, but aligning the WBS and the schedule network’s deliverable definitions means that the readers of the WBS and the readers of the schedule network are speaking the same language and that progress information can be trivially mapped back to the completion of WBS components. At project scale this requires automation support and the next article will describe the new œAdd and prune dependencies algorithm which facilitates this.