What triggered use of data analytics in production at BMW?
Smart data analytics is one of the main fields for digitalisation of our production system in addition to innovative automation & assistance systems, smart logistics and additive manufacturing. For all applications in our plants, the concrete benefits and the effect are always primary. For all digitisation fields, this means that if we see process improvement potential with a smart data analytics solution, we start a pilot project. If we achieve the desired results, we integrate this solution into the production process.
Where did the idea to analyse data come from?
A great data volume is required for the production of an automobile, and production and logistics processes also produce large amounts of data. The challenge is to determine which data you should consider to obtain relevant knowledge for improving your processes? Which data help to increase the availability of your systems? We consistently use the new possibilities to link and evaluate this data in an IoT platform to optimise our processes further. Maintenance is especially a great motivator; every minute less of plant downtime means that you can build one vehicle more without needing to adjust your capacities.
You use the term “smart data analytics.” What exactly makes them “smart” for you?
We believe that the term "big data" is rather unusual in the meantime. Collecting large amounts of data is little more than a first step. Finding out which data are worth analysing is somewhat more demanding. If you can also draw the correct conclusions, you are well on the way to a control cycle. You can not only optimise a very specific process, but also upstream and downstream steps.
Do you have an example of how a successful pilot project was implemented in normal operations?
An example is the fine control of presses by laser-marked body parts. Sheet thickness, sheet metal strength, texture of the surface and the degree of oiling differ between the sections (blanks) of one and the same coil. At our plant in Regensburg, each blank is provided its own ID by laser marking. As a result, we can fine-tune the presses to the properties of the blank, for example, additional oiling of the blank before forming by means of a control command. This reduces the number of rejected parts, and the material utilisation factor of a coil increases further. The benefits are so significant that we are going to use this smart data application in other press works in our production network over the coming months.
What experiences have you had in your pilot projects? Do you have sufficient sensor technology in production, or do you need to retrofit in this area?
In the highly automated production areas, the following applies: Robot and control technology are equipped as a matter of course with the necessary sensors. In assembly on the other hand, we have retrofitted conveyor systems with a large number of sensors, which in particular monitor temperature, vibrations, and electric power. Low-cost sensors are used, the favorable price-performance ratio of which enables widespread use. The systems transmit the data from these sensor kits and other process data via live stream into the BMW Intranet-of-Things platform, where they are visualised and analysed in real-time. Based on this data, our staff can assess maintenance requirements optimally.
What are the biggest advantages in general for production using smart data analytics?
Let me mention an important example with measurable results: Intelligent data management enables more accurate forecasts concerning wear and guarantees consistently high quality with manageable costs and effort. Smart data analytics provides completely new opportunities that go far beyond previous analysis possibilities. The speed, with which new solutions can be implemented, is increased significantly. For example, we have been able to achieve sustainable improvements for numerous predictive maintenance applications and significantly reduce planning times thanks to millimeter-precise digital maps of our plants.
Does this extend beyond predictive maintenance?
In addition to predictive maintenance solutions, we especially consider optimisation of diverse processes. The analysis of screw data helps us not only to monitor the quality of manual and automated screwdriving processes, but also to improve them in the long-term, for example, by including the findings in training sessions. With online process control, we monitor all automated manufacturing processes in the paintshops continually in the sense of predictive maintenance and best quality for our customers. In body manufacturing, we can optimise the installation of doors and closures data-based in regard to paintshop and assembly. Smart data analytics is therefore process know-how and core self-performance.
Are there sufficient experts for this in the market? Do you train them yourself?
Experts from different areas are required in the data analytics field. After all, not only data analysts analyse data, but above all people with experience in system integration and sensor technology. In this context, we rely strongly on our internal skills, in particular in maintenance. We train data analysts further as an area of competence.
See the video from BMW here