Efficient data management for the use of mathematical optimization models in production strategy

Global production networks enable manufacturing companies to be successful in a global market environment. A variety of methods are available for the design and coordination of these networks. The challenge in applying these methods in industrial practice is to provide up-to-date and high-quality input data. This challenge is to be met with efficient data management and the use of data analytics. 

Today, manufacturing companies face a global and volatile market. In order to adapt to these market requirements, respond to regional customer demands and reduce transport costs, more and more companies are relocating their originally centralized production to decentralized global production networks. Exploiting their potential is accompanied by increased complexity in planning and control.

In recent years, science has been able to develop a variety of methods to help companies cope with this complexity. The iwb also contributes to the scientific development by developing mathematical optimization models for this field of application.

However, all these methods require a large amount of input information and data to be executed. This data must not only be obtained once, but also updated regularly. In addition, the quality of the data has a significant influence on the results of the applied network planning methods.

At present, the collection, updating and validation of data requires a high manual effort and is mainly based on the assessment of experts which is fraught with uncertainty. These circumstances represent a major challenge for sustainable and efficient network planning.


The aim of the research project is therefore the development of an efficient data management for the planning of global production networks. On the one hand, the quality of the data is to be secured and increased and the manual effort for the acquisition, updating and validation of the data reduced.


The first step in the direction of efficient data management is the recording of data requirements and the classification of the required data. With the knowledge of the data requirements, strategies and methods can be developed in the second step to make data management more efficient.

One possibility for this is the use of data from various IT systems of the companies. Data on current production capacities, line capabilities, etc. can be derived, for example, from production data acquisition, whereby this process can be fully automated. Methods from the field of data analytics can be used to structure and analyse historical data, some of which is recorded in an unstructured form.

In addition, the validation of expert statements can be facilitated with the help of recorded historical data. For example, the ramp-ups of individual lines after rebuild can be viewed in the past and compared with the forecasts for the next rebuild.

Data sources outside the company can also be used to generate information. For forecasting cost rates for production and transport, data on the development of wage costs at the production sites or the forecast development of the oil price can also be taken into account.

In the final step, all methods are to be combined and prototypically implemented. In addition, it will be investigated to what extent manual efforts in data management could be reduced and the data quality increased.


Our thanks go to BMW AG for supporting this project.

Duration 01.04.2017 - 31.03.2020
Partner BMW AG