Digitalization projects relating to big data, smart data and the like provide companies with enormous advantages and increase their competitiveness. For example, companies can use predictive maintenance to plan maintenance times precisely and thus minimize machine downtimes. This is particularly efficient in terms of time and costs. The design and development departments can also draw conclusions from the data flowing back from production and optimize their processes. This is good for everyone involved - both the company and the customer.

Potential analyses for SMEs

Small and medium-sized enterprises (SMEs) often still find it difficult to assess whether, when and how big/smart data technologies are worthwhile for their own company. They often lack specialist knowledge and resources. It makes sense for them to bring external support on board for the planning and implementation of digitalization projects. SMEs can find this in special initiatives, funding programs and competence centers of the individual federal states, which offer local companies locational advantages in the use of Industry 4.0 technologies if required.

For example, the Smart Data Solution Center Baden-Württemberg, or SDSC-BW for short, provides SMEs with neutral and independent advice on smart data technologies with financial support from the Baden-Württemberg Ministry of Science, Research and the Arts (MWK). Launched in 2014 by Stuttgart-based Sicos BW and the Karlsruhe Institute of Technology (KIT), the Solution Center offers a free analysis of potential in particular. This analysis is carried out by KIT data analytics experts and gives companies an initial insight into the world of data analysis. They get to know the first smart data technologies and can then better assess how big and smart data can be used in their business environment. The only prerequisite: the interested company must compile and make the data available and appoint internal application experts to exchange ideas with the data analytics experts. Based on the results of the analysis, the SDSC-BW experts then provide recommendations for further concrete steps regarding possible fields of application and - where necessary - are available to assist with the implementation of a more in-depth project.

Many successful SDSC-BW project examples show that digitization projects can also be implemented in SMEs.

Project example: Better prediction of production time

Erdrich Umformtechnik, based in Renchen-Ulm, Germany, is a family-run company that has been supplying brake, chassis and drive parts for the automotive industry worldwide for more than 55 years. The high-quality and scheduled production of a large quantity of components is a complex process consisting of various individual steps. If one of these steps is delayed, the time lost must be compensated for elsewhere in order to keep to the plan. As the planning data available before the start of production is very limited, classic business intelligence (BI) tools reach their limits here. The aim of the project with SDCS-BW was therefore to intelligently enrich and evaluate this amount of data using smart data technologies.

The production of components requires several steps, some of which involve external suppliers. Different data is available for the various production steps, such as pressing, assembly and welding, for example from the machine used; at the same time, new data is added during production, including data on necessary machine repairs, machine speed and the number of units produced. Chassis components were selected for an initial potential assessment of the data; Erdrich provided the data for the entire production processes over a period of one and a half years.

The potential analysis began with a search for possible influencing factors for a planning delay. It quickly became apparent that certain machine workstations and component combinations lead to a deviation from the plan particularly frequently. The analysis raised the question of whether the planning approach used to date could react flexibly enough to the constantly changing machine conditions. The experts consulted with each other and extended their objective to testing dynamic planning. To this end, they enriched the data set with additional information from historical production data.

The result: the project team was able to identify various processing chains that can be used to improve the time estimation of production, in addition to important influencing variables that often lead to delays in production. In doing so, it evaluated a number of powerful and modern algorithms and potentially helpful exciting pre-processing steps. In addition to determining the typical statistical error measures, the results were also evaluated as cost differences from the original plan model and a very simple base model (mean estimator). In this way, the company learned that production costs can be predicted more accurately thanks to the improved production time estimation.

This gave Erdrich a new starting point for investigating possible production delays. With the help of the newly developed regression models, planning accuracy can be improved and dynamically adjusted. Even if machine data acquisition is not yet seamless, the results obtained will help decision-makers to make well-founded progress with the implementation of digitalization. The results can already be used to improve production time estimates.

Digitization projects (also) with small amounts of data

Practical examples like this illustrate this: It is a misconception that only large companies have the necessary data volumes to successfully implement digitization projects. Even if the term "big data" suggests it, the quantity of data is not the only thing that matters. Even SMEs such as Erdrich or much smaller companies have sufficient data that can be profitable directly or in combination with other external data or information - quality and variance are crucial. The data is of interest if patterns or connections can be identified in it that provide insights for process optimization. Smart data analyses are therefore not just about the volumes of data collected by IT, but also about combining it with other information - such as material properties or the experience of technicians. In this way, SMEs can also gain valuable insights from their data volumes and position themselves more competitively using smart industry and the like.

Dr. Andreas Wierse, Managing Director of Sicos BW / ag