Doctoral Dissertation: Development of a generic method for investigating temporal variability in production systems to improve production planning and control.

Generic data model to reveal optimization potential under consideration of the throughput

Motivation

The term "Industrial Internet of Things" was expressed for the first time in 2011 at the Hanover Fair. Today, the use of this term has already taken on inflationary features and many companies associate with this term currently more a marketing instrument. But the digital transformation of companies, which is to be approached in the sense of Industrial IoT, holds numerous optimization potentials. Flexibility in particular is becoming increasingly important in view of the inclined degree of product individualisation. Ensuring this flexibility sometimes requires additional production capacity. Companies are faced with the problem that a new machine would generate more output and more sales, but the machine would not be running to capacity. The payback period would be extended, which would have a negative impact on costs. Therefore it is important to start and to offer companies, by means of Industrial IoT solutions, possibilities to adapt their production with existing resources and to uncover unused capacity reserves. 

The multitude of events causes variability. Variability is therefore an inherent part of the systems, so we can speak of inherent variability. The result of this inherent variability is a reduction in the productivity of the production system, as the initial schedule cannot take all the events into account. The result is non-compliance with production program schedules due to excessive lead times or loss of throughput. In addition to production program planning, variability in production systems also influences the scheduling of deliveries, capacity planning or resource availability, inventory management, cost development, etc.

In order to take variability into account in production systems, it must first be measurable and quantifiable. In Six Sigma, it is said that processes cannot be optimized without measurable data. The more data we generate to analyze and monitor processes and their performance, the more knowledge we gain about them. This also applies to the influence of inherent variability, which is addressed by the doctoral project.

Objective

The aim of the doctoral project is to develop a method for incorporating the sum of events affecting a production system into the planning processes. This means that the method to be developed enables the investigation of variability in production systems. For ideal capacity forecasting and planning of the production system, every event must be taken into account and the degree of its influence must be factored in. Current approaches work with planning methods that take insufficient account of the dynamics of the production system caused by its complexity. The data basis is primarily based on rigid values. While the availability data of workstations changes dynamically over the production periods based on their downtimes, process data is usually measured and/or calculated and then stored as fixed values in the work plan. The inherent variability of production systems is identified as a missing calculation variable that needs to be recorded, measured and quantified. If the degree of influence of variability-causing events is known and the resulting variability can be quantified, this can be taken into account for future planning. The performance capacity of the system can then be realistically mapped so that planned and target values can be reliably set. This results in a high level of planning accuracy, which guarantees compliance with the production program plan and ensures that delivery deadlines are met. Meeting delivery deadlines has a positive effect on the service level. A positive service level is reflected in a high level of customer satisfaction, which is a decisive competitive factor.

Furthermore, identified causes of inherent variability can be minimized through targeted improvement measures, which increases system performance and promotes efficient use of resources. The control loop model of the generic method is shown in Fig. 2.