Application examples of Predictive Maintenance
Due to the Internet of Things (IoT), Big Data, the increasing networking and digitization of the business world and production, the field of maintenance and servicing has changed massively. The combination of all these technologies and their interplay provide an enormous database and now make it possible to make reliable forecasts and correct errors before they even occur.
The new possibilities and applications opened up by Smart Data, IoT and Predictive Analytics have already been the subject of the two preceding articles in this four-part series. In order to make this a little more tangible, a few concrete examples from the aerospace and transport sectors will be used here to illustrate the practical applications.
Predictive maintenance in aviation
One of the biggest financial problems of airlines is the cost they incur when flights are delayed or cancelled altogether due to mechanical problems. In the majority of cases, this not only affects one flight, but disrupts the entire process: Other flights may also have to be postponed, passengers booked on new flights or accommodated in hotels, crews cannot reach their connections, etc. Besides financial aspects, this also has a negative impact on the customer experience. Therefore, airlines have a special interest in anticipating part failures in time to avoid such delays and cancellations. An analysis of all relevant data makes it possible to make predictions about problems that might occur. The analysed data includes previously necessary repairs, anomalies during previous maintenance, serial faults or information on routes flown. By means of a complex, multi-stage algorithm based on this “historical” data, mechanical problems that will occur within the next 24 hours can be predicted. This enabled mechanics to carry out the appropriate maintenance or repair during regular service stops – and thus prevent unplanned outages.
Another example concerns the replacement of machine parts in aircraft engines. These are very sensitive and expensive pieces of equipment – but at the same time, replacing engine parts is one of the most common maintenance tasks. Data-based predictive maintenance and repair allows better planning of parts delivery, storage and replacement, so there is no waiting time. What needs to be analyzed for this purpose are mainly telemetry data collected by the large number of sensors in the aircraft. However, findings from previous maintenance work also play a role here in predicting future wear. In this specific case, the mechanics were provided with information as to which parts are to be replaced within the next month. This helps companies to reduce repair costs and to have exactly those parts available that will soon be needed for repair.
What applies to private aviation can also be transferred to other areas: In the Air Force, predictive maintenance can also save costs and optimize operating time. In the case of the Eurofighter Typhoon – the result of a project for the British Air Force – data-based predictive maintenance would save around 2.7 billion euros over 25 years – for the maintenance and repair of ten fighter aircraft alone. In this specific example, the aim was to use predictive models to identify potential part failure in good time, which is not easy because wear and tear depends on various external factors – including the climate at the aircraft’s main location. For the British Air Force, a total of 2000 parts for each aircraft were tested in a simulation: 40 megabytes of data per part formed the basis for the analysis model developed from this. The model calculates the effects of certain factors and changes in the environment. This makes it possible to predict when aircraft will fail for maintenance and repair – and to provide appropriate replacements. Maintenance itself is also optimized: In the case of the British Air Force, the number of failures has been significantly reduced because new parts could be ordered in time.
Predictive maintenance in rail transport
As already indicated in the previous text, the intelligent evaluation of data from IoT and Big Data can also be used for rail transport, in order to plan repairs better and obtain information about the life span of certain parts, similar to the example of aviation. Siemens is currently working on an “Internet of Trains” that will make train delays a thing of the past. Siemens is one of the largest international infrastructure providers in rail traffic. Using sensor data, big data and predictive analysis, Siemens wants to offer a customer almost 100% reliability. All possible factors are taken into account: From the temperature of the engine to the doors (open or closed), vibration and camera data from outside the train, but also weather data and other external factors affecting the train. On the one hand, this allows for predictive maintenance, on the other hand it increases energy efficiency.
For example, since the system was introduced last year on the Moscow – St. Petersburg line, there have been just nine delays (until May) – and that with 16 trains running back and forth several times a day. In the case of Deutsche Bahn, Siemens has been monitoring wheel bearings, gears and other relevant parts since October. So far, the system hasn’t missed a single part failure.
Other areas of application for Smart Data, IoT and Predictive Maintenance are the subject of the fourth and final blog text, which provides further concrete examples from other industrial sectors.