Have you ever heard about “predictive maintenance”? Do you have any idea what “FOC” stands for? Well, we can tell you it is closely linked to “big data”. Now you feel even more puzzled? Let’s knit together the pieces!
“Big data” is a frequently used buzzword, but a concept that is hard to grasp as it is subject to constant change. In principle, big data means large, diverse volumes of data from a very wide range of sources, which are created at high speed, often in real time.
Also at Zeppelin, we deal with a surging tide of data – and we use it to improve our products and services. Basically, we collect data, analyze it and ensure a good data management to achieve usable results. Collecting data only adds value when these data can be usefully assessed.
Big data as a prerequisite for predictive maintenance
Let’s have a look at one practical example:
Zeppelin sold around 107 generator sets for combined heat and power plants (CHP plants) in Germany. CHP plants are operated with gas engines using spark plugs (16-20 units, depending on the size of the power plant). If a spark plug fails before the next planned exchange interval, the entire power plant stops working – this happens multiple times within the course of any given year. Service contracts concluded with the customer specify that round-the-clock power plant operation with a minimum plant availability is guaranteed, so if a spark plug fails and the CHP plant stops, revenue falls and additional costs are incurred because service technicians need to be dispatched for unplanned service jobs.
Colleagues at the Power Systems business unit asked themselves whether the likelihood of failure and the remaining time before failure of spark plugs could be predicted. The objective was to significantly shorten response times and, through prompt scheduling and maintenance, reduce CHP plant downtime to a minimum. The analysis of various data streams (big data!) relating to breakdown, such as increasing ignition voltage and deviation of exhaust gas temperature, made it possible to determine a link to the spark plug failures. An interdisciplinary team at Zeppelin derived from this a model which predicts the likelihood of spark plug failure. This model – which de facto predicts when maintenance is necessary – enables Zeppelin to coordinate service better and earlier, with the result that costs can be kept low. The model has been implemented in a first step in one third of all CHP plants in Germany, and has already generated annual savings. After successful evaluation and optimizing of the systems, Zeppelin is in the process of implementing it in all remaining plants. With very similar methods, we are also able to predict when maintenance for huge ship engines is needed.
Linking predictive maintenance and FOC



But how do we manage all the data we collect in the frame of predictive maintenance and how do we manage the dispatch of the technicians? That’s where our FOC – which is short for “Fleet Operations Center” – enters the stage. Engines and systems can be linked to a data platform, developed especially by Zeppelin, on a non-manufacturer-specific basis, for the purpose of performing individual assessments of operating data. This enables the development of tools for early detection and intervention in the event of problems. Currently, all collected data are brought together in the FOC from approximately 298 connected marine engines as well as spark plug monitoring for presently 32 Cat engines that are used in combined heat and power plants (CHP plants). Helpdesk employees provide first-level support for customers and Zeppelin’s service divisions.
When it comes to the shipping industry, customers can enormously profit from this service: it’s easier to plan maintenance works and the time schedule for individual trips might be adjusted to the maintenance cycle.
There are plans to extend the platform to include additional applications and segments – for example locomotives – in the future. We Create Solutions!
Further information:
Zeppelin and Splunk – big data at the service of our customers (case study video clip)
[…] 2019: FOC – the way towards predictive maintenance […]