Project planning and management
If the purchase of components is delayed three weeks beyond the forecast, the probability of delivering the project in 20 weeks drops from 74 % to 37 %: purchasing is on the critical path, and every day it slips, everything slips.
01Scope and work breakdown
Planning starts by defining the scope: what the project delivers and what is left out. Then it is divided into ever smaller parts, down to tasks that one person can estimate and control: this is the work breakdown structure. For each task, its estimated duration is noted, along with the resources it uses —people, instruments, money— and which other tasks it depends on: the prototype cannot be assembled before the board exists, nor tested before it is assembled.
02Gantt and critical path
The Gantt chart, which was seen in the history of industrial organization, shows each task as a bar over time. The critical path method (CPM) calculates, with a forward pass, the earliest start of each task and the total duration; with a backward pass, the latest start that does not delay the end. The difference is the slack. The tasks with no slack form the critical path: any delay in them delays the whole project, and any effort to shorten it has to go there.
03Estimating with uncertainty: PERT
Durations are estimates. The PERT method asks for three per task —optimistic \( a \), most likely \( m \) and pessimistic \( b \)— and calculates an expected time and a variance. The project duration is approximated with a normal distribution, with the sum of the means and of the variances along the critical path:
Ten tasks, in weeks (a, m, b): A specification (1, 2, 3); B hardware design (2, 4, 8), after A; C firmware (2, 5, 10), after A; D purchasing (1, m, 8), after B; E board fabrication (1, 2, 4), after B; F assembly (1, 1, 2), after D and E; G integration (2, 4, 9), after C and F; H testing (1, 2, 4) and I report (1, 2, 3), after G; J presentation (0.5, 1, 1.5), after H and I. In red, the critical path; in light gray, the slack of each task.
04Resources, risks and monitoring
A plan without assigned resources is wishful thinking: if the same person designs the hardware and writes the firmware, those tasks are not parallel even if the chart says so. The risks identified in the technical analysis translate into time reserves and early actions, such as ordering long-lead components earlier. During the project, progress is made by milestones —approved specification, board working, integrated prototype, tests complete— and what has been done is compared with what was planned in order to correct in time. Agile methodologies propose short cycles with partial deliveries, useful above all in software. Responsibility for deadlines and for the quality of the work is part of professional practice.
05In the laboratory
Build the work breakdown structure of the final project down to tasks of one or two weeks.
Enter the tasks, the dependencies and the three estimates into a project management tool and identify the critical path.
Every two weeks, record the actual progress and compare it with the plan; decide what to do about the deviations.
06Common mistakes
- Estimating only the most likely case. Tasks slip more than they come in early; the pessimistic estimate matters.
- Forgetting component lead times. They are often on the critical path.
- Speeding up tasks that have slack. They do not shorten the project; the effort goes to the critical path.
- Planning without assigning people. Two parallel tasks of the same person are not parallel.
07Self-assessment
What is the expected time of a task with a = 2, m = 4 and b = 8?
\( (2 + 16 + 8)/6 \approx 4.33 \) weeks.
With the initial values of the lab, how long does the project take and what is the critical path?
About 18.7 weeks, through A, B, D, F, G, H and J.
What is the probability of finishing in 20 weeks?
Around 74 %.
What is the slack of a task?
How much it can slip without delaying the end of the project.
08Further reading
- Project Management Institute. A Guide to the Project Management Body of Knowledge (PMBOK Guide). PMI. (Spanish edition: “Guía de los fundamentos para la dirección de proyectos (Guía del PMBOK)”.) The reference framework for project management.
- Harold Kerzner. Project Management: A Systems Approach to Planning, Scheduling, and Controlling. Wiley. Project planning, scheduling and control in detail.
- Frederick P. Brooks. The Mythical Man-Month. Addison-Wesley, 1995 anniversary edition. Why adding people to a late project makes it later.