From manual condition monitoring to predictive maintenance

A production plant consists of hundreds or thousands of components which are responsible for smooth production. This might include robots, motors, and machine tools. The failure of a single component disturbs operations as a whole and reduces the efficiency of the plant. For this reason, the continuous monitoring of the condition of individual components is desirable.

With condition monitoring (CM), the foundation is laid for further applications such as digital twins, a smart factory, or predictive maintenance.

Some (component) manufacturers already presented CM solutions several years back, frequently with the integration of existing controllers (PLCs). However, these are almost always associated with significant initial costs, which means that it only makes sense to use them if the number of components is large. Other manufacturers continue to rely upon traditional interval-based maintenance.

Nonetheless, simple solutions with a good cost-benefit ratio are often available, too. One thing’s certain: A good overview of the technologies and the results that might be achieved is a prerequisite for success. It doesn’t always have to involve machine learning, a big IoT solution with a complex platform and vendor lock-in, or data storage abroad. A sense of proportion is key here.

In order to use condition monitoring successfully, we recommend that you examine the following aspects in advance.

Measuring machine values with condition monitoring

Do you have a problem that could be solved through making more information about the involved components available? Or a problem whose solution would benefit a customer so much that you could increase their profit margin? Then condition monitoring has a role to play. Single out the easiest and most lucrative application scenario for the initial implementation. The solution does not need to be particularly disruptive or original. A simple improvement in quality or increase in the OEE (overall equipment efficiency) is often enough. In this way, condition monitoring – by reducing failure rates – can return good results, including financially.

Typically, condition monitoring offers the following:

  • A nuanced understanding of the true condition of industrial machines, e.g. the mechanical factors and environmental conditions that result in machine downtime
  • Optimization of machine performance (OEE or speed, precision etc.)
  • Remote detection of problems
  • The early detection of changes to properties and parameters of the physical condition of an object or machine, possibly indicating the development of a fault. Such parameters include: Temperature, pressure, efficiency, process information, vibration, oil analysis, electrical check, noise emission, NDT (non-destructive testing)
  • Early detection of a potential machine fault and diagnosis of the problem, allowing measures to be implemented before performance problems or downtimes occur
  • Prevention of unnecessary, frequently costly maintenance
  • Employees are able to make better decisions, since their gut feelings and wealth of experience are supported by sensor data
  • Alerting
  • Machine simulations (prototype testing, model construction)

The use of condition monitoring results in (ROI):

  • Increased productivity (for example, because maintenance and repair work can be planned, thus optimizing the working window)
  • Development of a robust, condition-oriented servicing strategy for machines and equipment. Preventative and corrective measures can be planned for the optimum point in time
  • Improved processes and product quality
  • An assured innovative lead due to extra knowledge that can be used for prototype development

New opportunities for machine manufacturers:

  • New price models, e.g. pay-per-use instead of demanding an upfront fee for a machine
  • Improvement in product life cycle management

Architecture and technologies

Get an overview of your components and data and find out about processing opportunities. There’s no need to change your entire infrastructure. There are easy ways to collect data and integrate it into the everyday decision-making process. As a technician, for example, you could start familiarizing yourself with the IoT framework by looking at the Eclipse projects Milo, Ditto, and hawkBit. To process and display the data centrally, IoT platforms such as Thingsboard.io can be used. It’s likely that the data you need is already available in your environment, but in some cases, sensors need to be installed. As a rule, you cannot dispense with the generally central processing of data. Our advice to you: Don’t be discouraged by thoughts of a supposedly overwhelming architecture. Here, too, simplicity comes up trumps.

Normally, you must take the following infrastructure elements into account:

  • Network (wired or wireless): Receives context information from all machines or devices and ensures that they are all being monitored for abnormal behavior
  • IoT platform: Once the data has been retrieved from various different sources and protocols, it must be brought together, displayed, and interpreted
  • Saving data: Data can be saved in the cloud or at the local site (on-premise)

Simplified Architecture for Condition Monitoring.
Source: Own representation

Cyber security and data protection

Cyber security is particularly important when it comes to operational technology: On the one hand, the data reveals information about the core processes and on the other hand an attacker can take control of the devices if they are insufficiently secured. So who should get access to the data? Where is the data saved? Where is the data evaluated? Locally (on-premise) or in the cloud? How can you prevent incorrect data from being “smuggled” into the system in order to falsify the analysis or diagnosis? What links do the old machines use to transmit data? It makes sense to include security-relevant aspects such as access rights, network restrictions, and encryption during data transfer right from the initial prototype stage.

Condition monitoring of equipment.
Source: Own representation

Implementation

The business case is clear, the architecture and technologies have been selected, and the security requirements have been defined. Now comes the implementation. Once you’ve made sure that the machine can communicate with the right protocols in the desired way, you’ll see that some of your initial assumptions will need to be revised. This is because sometimes only real life can tell you which data is truly meaningful. So it’s important to get working on the real-life setup as quickly as possible. Consequently, you start with data acquisition and concentrate on the refinement of the app (so visualization, for example) only towards the end of the project. The focus is on corporate value and not on the technological “wow” factor. It’s important to keep your single business objective in mind.

Data collection is also important. In the case of unexpected downtime, you can use the data collected from the machine to determine which parameters changed before the problem arose and what caused the problem. In particular, this is important if the person performing the error analysis does not have the special knowledge required to find the cause independently.

An example: Condition monitoring for lifting platforms

Lifting platform with sensors.
Source: Own representation

Sensors detect measured values such as temperature, pressure, vibration, and oil level. Micro-controllers retrieve the measured values at regular intervals and transfer the data to the control system. An industrial gateway queries the measured values and passes them on to the processing system. This can be an IoT platform or a self-developed platform, and processing can be local (on-premise) or take place in the cloud.

The expectation is that the combination of the measured values allows conclusions to be drawn about the condition of a component. For example, changes in vibration behavior at the same time as a temperature increase might indicate insufficient lubrication even if the lubrication fill level does not seem abnormal. Once sufficient data is available, the dependencies between the individual factors become clear. Operating errors and material faults can be detected. If enough data is known, you can define threshold values that indicate the need for maintenance.

Analyzing machines efficiently through condition monitoring

For many, condition monitoring is their first step into the networked world of the IoT and Industry 4.0. To take this step, you don’t need to understand the large reference architectures or to implement expensive, ready-made solutions. Simply start out with the right mindset and you’ll soon have your processes and machine running times under control without investing too much time and money. This application scenario speeds up your digital transformation. It has excellent scaling potential and is compatible with future plans – it can even act as a preparatory step. Next steps might involve more extensive data analysis, predictive maintenance, and remote maintenance using AR technology.

A selection of our services:

  • Analyzing opportunities for your company, production, and services
  • Designing a solution that takes your digitalization strategy into account
  • Comparing and selecting technical components
  • Implementing the solution and embedding it in your system landscape

Let us help you learn about your opportunities in today’s brave new digital world. Here you can find out more about AR/VR and machine learning. You might also like to take a look at BPM. We’ll accompany you and support you during your journey!

Your direct contact

Novatec_Jonas-Grundler

Jonas Grundler

Head of IoT & Digital Innovation
Table of contents
Jonas Grundler Head of IoT & Digital Innovation
Novatec_Jonas-Grundler