Flow Validation
A well-constructed flow diagram ensures that material has a clear path from Input to Output, with no orphaned nodes or broken connections. This section covers how to verify your flow and track entities through it.
Structural Validation
A valid flow diagram satisfies these rules:
Every Flow Must Have Boundaries
- At least one Input node (material entry)
- At least one Output node (material exit)
- Without both, the flow has no defined start or end
Every Node Must Be Connected
- Each node should have at least one incoming edge (except Input nodes)
- Each node should have at least one outgoing edge (except Output nodes)
- Orphaned nodes โ not connected to anything โ indicate an incomplete route
Decision Nodes Need Multiple Outputs
- A Decision with only one output is equivalent to a simple pass-through โ it does not provide routing control
- Each output should represent a distinct condition
Merge Nodes Need Multiple Inputs
- A Merge with a single input is redundant โ it does not merge anything
- Merge nodes converge two or more paths
No Disconnected Subgraphs
- All nodes should be reachable from at least one Input node
- All nodes should have a path leading to at least one Output node
- A disconnected cluster of nodes is unreachable and has no effect on the material flow
Entity Tracking
Flow entities record their journey through the diagram:
History Trail
Every time an entity passes through a node, a history record is appended. The history includes:
- The node ID and type
- The timestamp of arrival
- Quality state at that point
This creates a complete log of the entity's path โ useful for traceability audits and root-cause analysis.
Quality Accumulation
As an entity traverses edges with quality impact, its quality flags are updated. Each flag records:
- The measurement or check that was performed
- The resulting quality level
By the time the entity reaches the Output node, its quality flags represent the cumulative quality assessment of the entire production path.
Characteristic Accumulation
Similarly, edges with characteristic impact append measurement data to the entity's characteristics record. This means:
- Dimension measurements from scan nodes
- Weight checks from functional scans
- Any domain-specific data recorded at checkpoints
The entity becomes a data carrier โ collecting quality and measurement information as it flows through the process.
Common Flow Problems
Dead-End Path
A Decision route that leads to a node with no further connections. The entity enters but has no way to continue. Fix by connecting the dead-end node to the next operation or to an Output node.
Infinite Loop
A rework cycle without an exit condition. If the Decision always routes to rework and never to the next operation, entities circle forever. Ensure at least one Decision output leads forward.
Missing Merge After Gateway
A Gateway splits flow into parallel paths, but the paths never rejoin. If synchronisation is needed before the next step, add a Merge node to converge the parallel paths.
Duplicate Edges
Two edges connecting the same pair of ports. This creates ambiguity in routing. Each port pair should have at most one connecting edge.
Best Practices
- Build the happy path first (Input โ operations โ Output with no branches)
- Add Decision nodes for quality gates after key operations
- Add rework loops only where the process actually permits rework
- Use Storage Points to explicitly mark buffers โ this makes cycle-time analysis more accurate
- Keep flow diagrams at a readable density โ split a very complex flow into multiple sub-diagrams attached to different operations
- Verify that every Decision node has at least one path that eventually reaches Output โ this guarantees flow completion