How do you draw a process flowchart?

Symbols and basic rules

Flowchart symbols only work if everyone reads them the same way. ISO 5807 is the standard that defines flowchart symbols. The ones used most are an oval for start and end, a rectangle for a process step, a diamond for a decision with two or more outcomes, and arrows for the direction of flow.

A few rules keep the chart readable. There should be one start. Every decision branch should be labelled. No path should end in an unexplained dead end, and loops should be drawn explicitly with their exit condition. If several roles are involved, a swimlane for each role makes responsibility clear.

Keep the level of detail consistent. Putting 'start the pump' next to 'replace the seal on valve 3' in one view confuses the reader.

Example: starting a pump

Here is a hypothetical pump start-up procedure, written one step per line:

• Start the pump.
• If pressure is above 4 bar, open the bypass valve.
• Otherwise, close the bypass valve.
• Log the pressure reading.
• End.

The pressure line becomes a diamond with two outgoing arrows, labelled yes and no. Both branches join at the step that logs the reading. The chart has one start, one end and one decision, so you can check it by walking each branch: at 3.5 bar the bypass should be closed, and at 4.5 bar it should be open.

Common mistakes

The most common mistake is drawing the process as it should be rather than as it is. Walk the chart through with an operator on the real machine and correct it.

The second is putting everything on one page. A chart with thirty boxes is rarely read, so split sub-processes into separate charts. The third is unlabelled decision branches and vague exits such as 'other cases', which leave the reader guessing.

The fourth is a loop drawn without an exit condition; in practice it describes a process that never finishes. Finally, updating a chart without a version number lets old and new versions get mixed up in the plant.

Follow-up questions