Data Analytics in Maintenance Planning – DAIMP
Övrigt - Project final report, 2019
Manufacturing industry plays a vital role in the society, which is evident in current discussions on industrialization agendas. Digitalization, the Industrial Internet of Things and their connections to sustainable production are identified as key enablers for increasing the number of jobs in Swedish industry. To implement digitalized manufacturing achieving high maintenance performance becomes utmost necessity. A substantial increase in systems availability is crucial to enable the expected levels of automation and autonomy in future production. Maintenance organizations needs to go from experiences based decision making in maintenance planning to using fact based decision making using Big Data analysis and data-driven decision support. Currently, there is lack of maintenance-oriented research based on empirical data, which hinders the increased use of engineering methods within the area.
The DAIMP project addresses the problem with insufficient availability and robustness in Swedish production systems. The main challenges include limited productivity, challenges in capability of introducing new products, and challenges in implement digital production. The DAIMP project connects data collection from a detailed machine level to system level analysis. DAIMP project aimed at reaching a system level analytics to detect critical equipment, differentiate maintenance planning and prioritize the most important equipment in real-time. Furthermore, maintenance organizations will also be supported in moving from descriptive statistics of historical data to predictive and prescriptive analytics.
The main goals of the project are:
Agreed data parameters and alarm structures for analyses and performance measures
Increased back-office maintenance planning using predictive and prescriptive analysis
Increased use of dynamic and data-driven criticality analysis
Increased prioritization of maintenance activities
The goals were further divided into specific goals and six work packages were designed to execute the project.
WP1 focused on the purchase phase and getting data structures and collaboration with equipment vendors correct from start.
WP2 focused on the ramp-up phase of new products and production lines when predictive and prescriptive analytics are important to handle unknown disturbances.
WP3 focused on the operational phase and to provide data-driven decision support for directing maintenance efforts to the critical equipment from a systems perspective.
WP4 focused on designing maintenance packages for different equipment with inputs from WP3, including both reactive, preventive, and improving activities.
WP5 focused on the evaluation and demonstration for different project results
WP6 focused on coordination project management
In WP1, models were developed to understand the missing element for the capability assessment from initiation of the machine tool procurement to the end of lifecycle. The information exchange and process of machine tool procurement from the end-users perspective was assessed. Additionally, the alarm structure is created using the capability framework and the ability model. In WP2, diagnostic, predictive and prescriptive algorithms were developed and validated. The algorithms were developed using manufacturing execution system (MES) data to provide system level decision making using data analytics. Improved quality of decisions by data-driven algorithms. Moved from experienced based decision to algorithmic based decisions. Identified the required amount data sets for developing machine learning algorithm. In WP3, data-driven machine criticality assessment framework was developed and validated. MES and computerised maintenance management system (CMMS) data were used to assess criticality of machines. It serves as data-driven decision support for maintenance planning and prioritization. It provided guidelines to achieve systems perspective in maintenance organization. In WP4, a component classification was developed. It provides guidelines for designing preventive maintenance programs based on the machine criticality. It uses CMMS data for component classification. In WP6, three demonstrator cases were performed at (i) Volvo Cars focusing on system level decision support at ramp up phase, (ii) Volvo GTO focusing on global standardization and (iii) a test-bed demo of data-driven criticality assessment at Chalmers. Lastly, as part of WP6, an international evaluation was conducted by inviting two visiting professors.
The outcomes of the DAIMP project showed a strong contribution to research and manufacturing industry alike. Particularly, the project created a strong impact and awareness regarding the value maintenance possess in the manufacturing companies. It showed that maintenance will have a key role in enabling industrial digitalization. The project put the maintenance research back on the national agenda. For example, the project produced world-leading level in MES data analytics research; it showed how maintenance can contribute to productivity increase, thereby changing the mind-set from narrow-focused to having an enlarged-focus; showed how to work with component level problems to working with vendors and end-users.