How do I accurately track the amount of downtime caused by a changeover?

Here is how to use ArchFX React to accurately measure the duration of a downtime caused by a changeover.

1. The operator stops the existing job, starts the next job, and takes the line down for changeover and React detects a downtime.
2. React automatically detects a downtime after the downtime detection threshold for the new product is reached.
3. The operator labels this downtime as due to the changeover.

When the first unit of product in the next job passes the React pacemaker machine, the downtime will automatically be ended and the amount of downtime due to the changeover will have been accurately measured. 

The amount of downtime will be accurately measured even if the downtime crosses a shift change event. Let's say the changeover downtime begins two minutes before the end of shift and continues 13 minutes into the next shift. React will automatically "split" the downtime in its internal data model at the shift change (so React can divide downtime amounts between shifts, etc.), but it will copy all the metadata from the first part of the downtime on to the second part, so in the parts of the UI and analytics that are not specific to a single shift, this will appear as a single 15-minute long downtime due to changeover. The logical "splitting" of the downtime event is only relevant when you're trying to look at data for a single shift.

Later, if you want to export the data about the changeover downtimes to a CSV, you can use the "React Download CSV" feature under Production Insights to export the recorded downtime data (which of course includes the changeover-related downtimes) to a CSV. Finally, you just use standard filtering functionality in Excel or other spreadsheet software of your choice to filter and show only the changeover-related downtime data. 

As always, if the behavior you see doesn't match this specification, please click "Report Issue" when you next see the problem and also please email help@archsys.io a bug report with full job name / timestamp / timezone information for where you saw the problem occur so we can investigate.