Skip to content
Project Management | Part 5 of 10

Waterfall, Agile, Hybrid: Choose per Workstream

Plan driven, iterative and flow approaches each fit a different kind of work. Mechatronic projects need all three, connected at fixed points.

In many companies the question “waterfall or agile?” is answered by belief. The software team wants sprints, the plant engineering team wants gates, and each side thinks the other is doing it wrong.

Both can be right. The approach should follow the nature of the work, and in an industrial project the work is not all of one kind.

Three approaches, three kinds of work

  • Plan driven, waterfall or V-model: requirements are stable and testable, parts are physical, changes after the design freeze are expensive. Core tools: work breakdown structure, critical path, gates.
  • Iterative, Scrum: requirements change or are unclear, results can be shown every two to four weeks. Typical for software and user interfaces. Core tools: backlog, sprints, reviews.
  • Flow, Kanban: a stream of similar items such as change requests, tests or drawings. Core tools: board, work in progress limits, lead time.

Five questions

Ask them for each workstream, not for the project as a whole:

  1. Are the requirements stable and testable today? Yes: plan driven.
  2. Is a change after the design freeze very expensive? Yes: plan driven with gates.
  3. Can a usable increment be shown every two to four weeks? Yes: iterative.
  4. Is the work a stream of similar items? Yes: Kanban.
  5. Do hardware and software develop in parallel? Yes: hybrid.

The hybrid rule

A new assembly cell is a typical hybrid project. Mechanics and procurement follow the schedule, because a steel frame cannot be delivered in increments and drive units have twelve weeks of lead time. The PLC and operator interface team works in two-week sprints with its own backlog.

The two worlds meet at fixed integration points, which are milestones in the master schedule:

  • Interface list frozen: every signal between mechanics, electrics and software agreed.
  • Software on the simulation model: tested before the real cell exists.
  • Software on the real cell: commissioning starts with a released version.

The mechanical lead attends the sprint review before each integration point. The product owner attends the design reviews. That is all the coordination most hybrid projects need, as long as it actually happens.

What goes wrong

The typical failure of a hybrid project is two plans that do not talk. The schedule shows “software” as one bar of ten weeks, the backlog has no dates at all, and nobody notices that the interface list changed three times after the freeze.

Fix the integration points, put them into both plans, and treat a missed one like a missed milestone on the critical path. The method war ends when both sides see the same three dates.

Newsletter

One Lean idea.
Every week.