Workshop tools

Two small Python apps for two shop-floor methods: OEE and time study.

Each one reads a CSV file or its sample data, then goes from the calculation to a decision: where to act first for OEE, how many operators for the time study.

The calculations are kept apart from the interface and covered by tests; both interfaces are in English and French. The examples on this page use the sample data that ships with each repository.

Tools
  • Two Python apps
  • Streamlit
  • pandas
  • Plotly

OEE dashboard

You load one CSV of production and downtime. The app checks the columns and values, then works out availability, performance, quality and OEE by machine, by shift and by day.

A downtime Pareto, adjustable targets and a downtime-reduction scenario help choose the first action; the filtered tables export to CSV.

OEE app, overall performance: availability 90.1%, performance 95.7%, quality 96.8%, OEE 83.4%, scrap rate 3.2% and 380 minutes of lost production, each against its target, then OEE and scrap rate over three days
The overview, on the sample data that ships with the app: each indicator against its target.
  1. OEE app: the production file upload, the sample data ticked, then the analysis scope, 8 rows, 2 machines, 2 shifts and 5 downtime reasons

    1Load the production data

    One CSV per workshop, or the sample data. Columns and values are checked before anything is worked out.

  2. Downtime losses: Pareto of downtime by reason with the cumulative curve, downtime by shift, then the table of top reasons

    2Read the losses

    The downtime Pareto by reason and the split by shift: breakdowns are 39.5% of the sample’s downtime.

  3. Action plan: a 20% breakdown reduction scenario, projected OEE 84.1%, availability 90.9% and 30 minutes of downtime saved

    3Choose the action

    A scenario cuts one downtime reason and works the OEE out again: 20% fewer breakdowns take the sample from 83.4% to 84.1%.

Sample result

The sample data that ships with the app: 8 rows, 2 machines, 2 shifts, 3 days.

  • Availability90.1%
  • Performance95.7%
  • Quality96.8%
  • OEE83.4%

The formulas

Availability
Operating time ÷ planned time
Performance
Ideal cycle time × units produced ÷ operating time
Quality
Good units ÷ units produced
OEE
Availability × performance × quality

Where to act first

Availability below 85%
Cut downtime
Performance below 90%
Look for speed losses, micro-stops or a wrong ideal cycle time
Quality below 95%
Work on the causes of scrap and rework
One reason above 40% of downtime
Deal with it first

Time study

You enter the observed cycle times or load a CSV file. The app works out the normal time with the performance rating, then the standard time with allowances.

It then gives the capacity per operator, the takt time and the staff needed, flags outliers (the 1.5 × IQR rule) and estimates how many readings to take.

Time study app, results: mean observed time 53.00 s, normal time 58.30 s, standard time 67.05 s, capacity 375.9 units per shift, 2 operators required, takt time 63.00 s, then an improvement suggestion
The results of the repository’s example: the standard time is above the takt time, and the app says so and suggests what to look at.
  1. Observed cycle times entered, 52, 55, 51, 54 and 53 seconds, then the outlier check: none found

    1Record the times

    Cycles are typed into a table or loaded from a CSV; every outlier is flagged.

  2. Chart of observed cycle times around the 53-second mean and below the upper reference limit

    2See the variation

    The readings against their mean, their distribution and their spread.

  3. Capacity and labour scenarios: total capacity, gap and utilisation for one to six operators

    3Size the team

    Capacity for one to six operators: with two, the demand of 400 units a day is covered.

Sample result

Five cycles observed: 52, 55, 51, 54 and 53 s. A 480 min shift, 60 min of breaks, 400 units a day.

Mean observed time
53.00 s
Rating
110%
Normal time
58.30 s
Allowances
15%
Standard time
67.05 s
Capacity per operator
375 units per shift
Takt time
63.00 s
Staff
2 operators (1.06 worked out)

The formulas

Normal time
Mean observed time × rating
Standard time
Normal time × (1 + allowances)
Capacity
Available time ÷ standard time
Staff
Demand × standard time ÷ available time
Takt time
Available time ÷ daily demand