Hi @TIM
This is a classic Event Analytics use case, start from your cycle search (the "Cooking Cycle" events), open Event Analytics, and add one calculation per suspect: avg steam pressure, avg steam-valve position, max product level, initial product temperature (start-of-cycle value), and cooling-stage duration.
Then use the three views in sequence — it's the community-recommended refinement triple:
- Histogram of cycle duration to see whether long cycles are a distinct population.
- Parallel coordinates brush the long-duration cycles and watch which factor axes separate with them. That visually nominates the culprits.
- Scatter duration vs. the top suspect, with a regression line, to quantify the relationship.
Color-code by recipe/product if you run more than one, and filter startup/shutdown noise at the VBS level so the distributions stay clean.
Two resources that walk this through end-to-end: the Getting to the bottom of bad batches write-up, and the Event Analytics in Action VILT recording, which covers comparing performance across runs and turning the patterns into actionable insight.
Kind regards
Frederik
