Analysis of Historical Data and Development of a Digital Product for Maintenance Optimisation

Abstract

This thesis describes the development of a digital tool that digests, analyses and displays industrial maintenance data with the aim of facilitating decision-making in maintenance scheduling. The initial datasets are a combination of CSV files where maintenance jobs are represented as rows, and each column includes specific information of the jobs such as the maintained machine, the job type, and machine location, among others. The process of developing this product started with a data exploration phase, which evolved into a calculation of different metrics, KPIs and benchmarks that would help understand the status of maintenance in industrial plants. The definition of such metrics, KPIs and benchmarks was performed iteratively with constant communication with Product Managers, Data Scientists, internal stakeholders and customers. The overall result of this thesis is, firstly, a list of metrics, KPIs and benchmarks that optimally describe maintenance schedules in industrial plants, and secondly, a web-based dashboard that displays this information depending on user’s interaction. From the customers’ point of view, the use of this tool allows them to quickly identify maintenance hotspots among their machinery fleet, visualise underlying maintenance trends, and compare current performance to different benchmarks, including industry averages. As a direct consequence of analysing this information, customers are able to take data-driven decisions to optimise their maintenance schedules within their plants