Search

Logistics & Manufacturing

DESIGN THINKING

Unveil approach

Industrie 4.0
PREDICTIVE MAINTENANCE

THE CASE OF PREDICTIVE MAINTENANCE

The Big Data Challenge

The amounts of data generated by a highly complex industrial plant are tremendous. This data is commonly characterized by the 5Vs.

SELECT AND DISCOVER THEM ALL!

Variety Variety

Variety

Value Value

Value

Veracity

Velocity

Volume

×

The Big Data Challenge

The amounts of data generated by a highly complex industrial plant are tremendous. This data is commonly characterized by the 5Vs.

SELECT AND DISCOVER THEM ALL!

Variety

The integration of different data sources as well as the need to integrate unstructured data are challenges that traditional relational databases are hardly able to handle. Especially because the system has to be flexible enough to integrate yet unforeseen data sources in the future. In the predictive maintenance context, sound, video, radar data, images etc. can be used in combination with other sensor data.

Value

Data in itself contains no value - the value is generated by using it. Determining the right variables from the massive amounts of redundant bycatch and the refinement of value-adding models are among the major challenges in any predictive maintenance project.

Veracity

Not all data is true, and sources can deceive you. The challenge with big data is to develop ways to put the good in the pot, the bad in the crop. Measurement errors e.g. have to be detected and eliminated based on models built on historical data.

Velocity

Production processes perform in real-time. The data collected is thus of high temporal granularity and requires real-time processing, if it should serve for adjusting ongoing processes as soon as parameters reach critical levels.

Volume

Large scale sensor area networks in industrial plants, controller area networks in vehicles produce vast amounts of data in dimensions of zettabytes, even brontobytes. These amounts of data are too large to store and process with the means of traditional database models. Using distributed systems (like Hadoop), databases are becoming networks.

DATA SOURCES AND INFRASTRUCTURE

The goal is to ensure that the IT Infrastructure actually meets the demands of the envisioned goal. A major challenge is the extension of classical, relational databases through non-structured data.

infrastructure

The choice of the architecture layout primarily hinges on how fast generated data needs to be turned into action via the step of analytics and which degree of process automation is to be achieved. This question, as the choice of relevant data sources cannot be generalized and highly depends on the (industry) specific use case.

DATA EXPLORATION AND MODELING

Goals and possibilities are evaluated from various business, technical, legal and IT aspects. Experts from the various fields have to convene and develop new ideas. Once the goals are set, data scientists develop statistical models that define and integrate all variables to predict when a failure of a component of an engine or machine will occur.


The models are then tested with training and real available data: this allows assessing the quality of the model, which will be refined further.

 
 
 
 

PROCESS INTEGRATION

 
 
 
 

Predictions and analytics results need to be embedded in the company's processes. Relevant persons have timely access with the right tools and a user interface that supports their decision finding.


The choice of the right tools depends on the desired degree of automation, i.e. in how far data is further analyzed and interpreted by the company’s staff and whether the model triggers the execution or issuing of maintenance tasks automatically.

While from the IT perspective Predictive Maintenance is at its core a Big Data challenge, it requires expertise in the entire variety of technological fields: network and telecommunication, IT architecture, the IoT, Cloud Computing and security.

REPLY HAS A PROVEN TRACK RECORD IN ALL AREAS THAT ARE RELEVANT TO PREDICTIVE MAINTENANCE.

Reply actively contributes to the development of standards for the IoT as a prerequisite for the development of Industrie 4.0 and predictive maintenance through its engagement in the OPC Foundation, creating the interoperability standard for industrial automation.

Reply's experience and technical expertise gained in numerous projects is helping to open new horizons in promoting Industrie 4.0, and in this context, Predictive Maintenance has to be regarded as a key accelerator.

PREDICTIVE MAINTENANCE IN PRACTICE

WIND
FARMS

COMMERCIAL
VEHICLES

CONNECTED
CARS

STEEL
PRODUCTION

Wind farms

Wind energy is one of the typical examples for predictive maintenance. The use of sensors connected via internet can almost completely eliminate the losses through unplanned downtimes of wind turbines. Especially for turbines in off-shore wind parks, remote sensoring in combination with predictive models help to reduce the need for routine checks on site and thus lower maintenance costs significantly.

Commercial vehicles

Predictive maintenance based on sensor data helps to prevent production downtimes on e.g. construction sites or in the agricultural sector and reduces accidents for instance in the logistics industry. Providers of rental commercial vehicles can dispatch service teams just in time, before a vehicle breaks down and even develop innovative business models like pay-per-performance.

Connected cars

Through the use of internal connected sensors connected cars are able to determine, whether one of the systems is prone to malfunctioning or failure. The data generated in e.g. the controller area network can be transmitted for further analyzing and a reminder can be triggered in the console to schedule an inspection. Information about the car’s condition can be transmitted to the workshop, so the necessary parts and materials are available and repairs can start without delay.

Steel production

Following the Design Thinking approach, it was possible, to detect that in a steel production company, sensors were already collecting all kinds of data generated by the production process. Designing a set up using IoT-Technology and IT-Integration by integrating the experience and know-how of the engineers, enabled the company to actively use this data to make predictions about material quality and eliminate flaws during the process significantly lowering the costs for later repairs.

 
 
 

Syskoplan Reply is a Reply Group company that focuses on consulting services and implementation of SAP technology. Syskoplan Reply partners with its clients to introduce innovative solutions in support of processes, thanks to its strong emphasis on SAP HANA, the Internet of Things, real-time analytics on the HANA platform and the new user experience with SAP Fiori. This approach is the point of excellence of Syskoplan Reply and main benchmark for companies that intend to integrate Customer Engagement and Commerce, digital solutions and Customer-Supplier centric models.