Högskolan i Skövde

Discrete Event Simulation (DES) has been a powerful tool for modeling and optimization of production lines for many years. Developing detailed models may be very time consuming and might also be computationally costly to execute, especially if optimization techniques are to be applied. Aggregation techniques, simplifying a detailed system into fewer objects, can be an effective method to reduce the required computational resources as well as a way to shorten the development time. An aggregated model can be used to identify the main constraints in a system, dimensioning inter-line buffers, and focus development activities on the critical issues from a system performance perspective.

An aggregated model of a real-world plant wide production system has already been developed in the discrete event simulation software FACTS Analyzer. Due to the limitations of FACTS Analyzer this model was instead rebuilt using the discrete event simulation software Plant Simulation which, while less user friendly, has much more configuration options. This model is then validated against the FACTS Analyzer model to ensure accurate results that can be implemented on a real life system.

In addition some typical optimization experiments are carried out on the Plant Simulation model and the results analyzed. The first of these optimizations alters the buffer sizes to find the minimum amount of buffer capacity needed to get the minimum lead time while maintaining the highest possible throughput. The second optimization uses the same parameters but also adds alteration of the batch size. Both of these optimizations shows that it is possible to get better results with lower settings on both the buffer capacity and batch size then the default value.

At the end of the report the project is evaluated and conclusions are drawn on what can be improved in the model, such as adding more streamlined updating of variant sequence.  Additionally future experiments that can be interesting to look at is specified based upon the results from the project.  

This thesis work in automation technology at 30 credits is conducted at Volvo Cars Cooperation (VCC) in collaboration with the University of Skövde.

With the current staffing method VCC are experiencing difficulties achieving an optimal utilization of operators as there has been difficult to ascertain how much work an operator can handle. Operators are an often overlooked part of simulation studies because they are complex to model. The completed thesis consists of developing a standardized operator object in the simulation software Plant Simulation, developed by Siemens. Using this object, it will be easier to staff optimally with respect to the number of work areas, operator skills and how many operators that are required for the given work areas.

The existing VCC Plant Simulation library already has an object for operators. This operator object was used as the basis for the development of the new operator object. When developing the new object, data collection was made in form of both qualitative and quantitative data to find out what functionality the new object needed. The existing object was evaluated against the functionality requirements to determine what were needed to be added, removed or modified. Five main areas were considered important to develop: (1) operators that can work across two or more work areas, (2) walking time for operators, (3) learning effect, (4) the learning curve for operators and (5) operators who work on different shifts. The learning curve was only partly developed, but the other four areas were fully implemented.

In order to test and analyze the object, three experiments were carried out using different case studies, where one of these cases studies was a real production line. The experimental design was jointly created with the VCC reference group and the University of Skövde. The main objectives of the experiments were to reduce the number of operators and increase throughput per hour. The result of the optimisation of the real production line showed that a significantly smaller staffing could be achieved without a major effect on the throughput.

The results of the project were that a standardized operator object was developed according to the specifications and objectives set for the project. This object provides the ability to quickly and easily implement operators in a simulation model. To verify the applicability of the operator object, it was successfully imported and used for simulation-based optimisation in three different simulation models. The operator object also contains functionalities that provide more realistic operators, such as the walk time to the operations, operators that can work across several work areas and competence for operators as well as learning.

The results of this project involve a simpler implementation of more realistic operators and thus lead to more realistic simulation models.