Weibull reliability study and replacement intervals for mine haul trucks

Karoo's 34 haul trucks each move about 220 t a load. When a truck stops unexpectedly in the pit, the mine loses roughly R38,000 of production an hour until it is recovered or replaced. Last year unplanned failures of three components caused 61% of unplanned truck downtime: - turbochargers on the diesel engines; - front suspension struts; - final drives, the last gear reduction at each rear wheel.

The maintenance manager wants to move from "run to failure" to planned replacement where it makes sense. The reliability engineer left, and the team has only a CMMS export of removal records. Some components failed. Others were removed while still working, during planned overhauls, and nobody knows how to use those records.

A sister operation in the Alberta oil sands, with the same truck model, wants the same analysis.

Your job is to: - build a dataset of failures and suspensions for the three components from the generation rules in task 1, since real CMMS data is confidential; - fit two-parameter Weibull models properly, including suspensions, with two methods and confidence bounds; - interpret the shape parameter to decide whether each component is wearing out, failing at random or failing early; - calculate a cost-optimal age-replacement interval for each component where planned replacement makes sense; - forecast the number of failures and spares needed next year.

You do not need mine access. The work is done in Python with SciPy, or with the open-source reliability library.

Done looks like Weibull parameters with confidence bounds for each component, a replacement policy with intervals in engine hours and the expected saving in rand per year, a 12-month spares forecast, and a short report for the maintenance manager that also works for the Alberta team.