Production run analysis

Compare machine variables across several production runs and time ranges, as time series, distribution, correlation, statistics or run comparison.

Production run analysis overlays several production runs from the same machine so you can see where they differ. The typical question: one run went well, the next went badly, and you want to know which process parameter changed.

Two variables across three production runs, overlaid as time series

Prerequisites

• A machine with active variables

• Production runs booked on that machine

How it works

1

Select site and machine

In the sidebar. Every list below refers to this machine.

2

Select variables

Under Variables, tick the variables you want to compare. The table shows Name, Unit and Type; the filter fields under the column headers help in long lists.

Selected variables appear as chips below the table. Load group pulls in a saved selection from Variable groups.

3

Set the time ranges

Under Time Ranges, choose between two modes:

• Orders: tick production runs in the table. Each run becomes one comparison range.

• Custom Time Range: enter start and end freely, independent of runs.

Additional column adds an extra run column, useful for filtering on an attribute that isn't shown by default.

4

Load the selection

Select Load Selection at the top right. Nothing is fetched until you do.

The five views

Five tabs sit over the same selection.

The Order Comparison tab puts the selected runs' metrics side by side

Tab	What you see
Time Series	One plot per variable, one line per range, all normalised to a shared duration so runs of different lengths line up.
Distribution	One boxplot per variable, one box per range. Spread rather than progression.
Correlation	Scatter plot of two variables against each other, points coloured per range. Pick the two variables from the dropdowns above the plot.
Statistics	Mean per numeric variable and range, plus the percentage difference between them.
Order Comparison	Metrics per run: duration, average throughput, quantity, availability, performance and productivity.

Order Comparison stays empty in Custom Time Range mode. Metrics need a run to attach to.

Statistics cover production time only

One difference that catches people out:

Statistics, Distribution and Correlation cover actual production time only. Samples inside detected downtimes are excluded. Time Series and the CSV export show the raw data, downtimes included.

The same throughput can therefore have a higher mean under Statistics than the plot under Time Series suggests. That is deliberate: a mean that counts downtime zeros does not describe how the machine performed.

The logic matches how availability is calculated for OEE. See Availability.

Exporting data

Below the plots is a CSV download for the loaded selection. The export contains raw data, so downtime periods are included. For larger or recurring exports, use Data export.

Common questions

Can I compare variables from several machines?

No. One analysis covers exactly one machine. To compare across machines, export the metrics via Data export and combine them elsewhere, for example in Power BI or Grafana.

Why are the plots empty when runs are selected?

Check three things: is at least one variable ticked, did you select Load Selection, and does the variable return values in that range at all. Variables of type STRING do not appear in numeric views.

What does normalising to a shared duration mean?

The x-axis under Time Series defaults to Duration: a four-hour run and a twenty-hour run both start at zero so their progression can be compared.

Switch the axis to Time and normalisation is dropped, showing the runs one after another with their real timestamps.

Related topics

• Product analysis answers the same question across every run of a product rather than a hand-picked few.

• Variable groups if you need the same variable selection regularly.