Requirements & Enabled Improvements Related to the Implementation of Advanced Preventive Maintenance Methods

Abstract: Currently well established technology-based manufacturing companies with a service business provides their markets with innovative and reliable products. Despite this, these companies often find their service delivery process to be inefficient, due to its time-based nature. Furthermore, the markets on which they operate are currently experiencing a shortage of qualified service personnel and thus sees a need to optimize the existing service delivery process. The collection of IoT data is an early step in the implementation of a more effective service delivery process. This data can be analysed in order to provide insight of the underling factors of machine breakdowns. If the company understands these factors they are also able to counteract the downtime related to failures and create new opportunities for the company. The purpose of this thesis is to investigate the requirements and possible effects related to the implementation of preventive maintenance methods based on quantitative analysis of Internet of Things data, gathered from connected units in an industry environment, for a technology based manufacturing company with a service business. This thesis found that predictive maintenance is the best suited advanced method for technological-based manufacturing companies with a service business and characteristics similar to the ones of the case company. Through implementing advanced preventive maintenance methods the service delivery process will grow more technologically dependant. Furthermore, it will render the conventional pricing system useless and favour a new performance-based pricing system for service delivery. Additionally the implementing companies will experience a reduction in the need of educated service technicians. Which arises from a reduction of reactive service visits related to the implementation of advanced preventive maintenance methods. This thesis was not able to construct a reliable failure prediction model for the case company, yet it identified several previous examples and methods where it has been done successfully. Apparent is that in order to archive sufficient results a considerable amount of domain knowledge is needed. Complementary data, which describes the setting of the predictions is also necessary. Further, there are heavy requirements on the data being analysed, both with regard to quality of measurements, sample size and in particular the length of the studied time interval. Furthermore, this thesis found that customers lacks a strategic relationship with their device. The far majority of errors is caused by external factors inflicted by customers.