The key to the success of IoT in maintenance is to start collecting data as soon as possible.

--In Part 1, we asked about the potential for IoT-based maintenance work. Are companies and factories increasingly implementing IoT now?

The use of IoT is a frontier area in the field of maintenance, and more factories are implementing it. There are also growing numbers of IoT tools themselves that support maintenance. For example, a tool to monitor vibration by attaching sensors to the drive parts of factory equipment, such as motors, is an easy example. In addition, there are systems that not only measure vibrations, but also perform frequency decomposition and other data processing, and if an abnormality is detected, the computer will sound an alarm. These systems are for the purpose of predictive maintenance that I mentioned in Part 1.

There is also a more rudimentary form of IoT. For example, pieces of equipment are linked to each other through communication so that if a problem occurs in one area, the equipment in the next process will automatically stop. This is classified as reactive maintenance, where a failure is dealt with after it has occurred, but since the failure can be detected quickly, the impact on subsequent processes can be minimized. If you notice a failure 10 seconds too late, it could also cause a machine to fail further down the line.

Another example would be a machine with an arm that grabs a certain part, and if for some reason it fails to grab the part, the line automatically stops. Another common example is preventive maintenance when a part that should not flow through the line passes through, the line is stopped to prevent accidents and breakdowns because the part could damage equipment further down the line.

--Is it possible that the use of IoT will allow for more standardization of work that used to rely on the intuition and unique skills of maintenance technicians?

Yes, it is. I think there is enormous potential, in particular for preventive and predictive maintenance where equipment data is monitored. It is also important for companies and factories that have no previous experience in implementing IoT to start small and be flexible in their implementation.

There may be cases where suddenly implementing IoT in the entire factory will not work. Therefore, it is advisable to first try to implement IoT in a limited way, such as by implementing it partially or setting up model lines or model equipment. Of course, it is assumed that you have a picture of what implementation in the entire factory should look like, but it is wise to start with a partial, small start and then expand the scale of implementation once you have gained the expertise.

--If you start small, you will be able to implement IoT quickly and easily.

Implementation as soon as possible is also extremely important. That’s because IoT-based maintenance will not be able to reach its full potential unless more data is collected and accumulated. If IoT is used for reactive maintenance to notify that equipment has stopped, it can provide immediate results, but for preventive or predictive maintenance, the obtained data must be analyzed and technical evaluation must be added.

The number of IoT maintenance tools is increasing, and more factories are adopting them, but in fact, there are many areas where the detailed methodology has not been completely established, such as where sensors can be put to be more effective, what types of data must be obtained to predict failure, and how obtained data should be analyzed. No one yet knows the optimal solution for IoT-based maintenance.

This is where it becomes important to collect as much data as possible and discuss internally the best way to use this data. To have an accurate fact-based discussion, it is important to have as much data as possible to use as raw material.

--So starting small is effective for collecting as much data as possible.

That's right. I don't think critical failures of equipment happen that often. It is important, then, to collect data early on to understand exactly what kind of behavior will lead to what kind of failure, and what the trends are.

Moreover, as I mentioned earlier, no one yet has the best answer to the question of what sensors should be installed on what equipment and what data should be collected to predict failures. Consequently, this means that the companies and factories that quickly gain this knowledge will have a competitive advantage. Starting small is also an effective way to develop your company in the future. Another important thing to do then is to analyze the data you have collected.

Will the future of maintenance be a data-based approach in addition to a hands-on approach?

--Data analysis is an area that has not been a part of maintenance work in the past.

And yet, it is a key feature of IoT-based maintenance work. IoT may bring about a change from the traditional hands-on approach of maintenance work that relied on intuition and unique skills to a hands-on approach that relies on data. This change will also greatly enhance the quality of maintenance. In other words, in the past, maintenance work has been handled by a limited number of on-site people, but now a new way of conducting maintenance will be required that involves people like data scientists who can analyze data in detail.

In order to achieve this, it will be important to acquire personnel who can accurately analyze the collected data, and it will also be important to determine how this staff can work with the on-site maintenance technicians. To establish this type of internal collaboration, those at the top must also reaffirm the importance of maintenance and redefine its value for the entire company.

--It seems that the task of analyzing the data obtained from IoT will be just as important as implementing IoT.

Even if you use IoT to store plenty of data, if you can’t make use of it, it’s useless. For example, even if you obtain the vibration data of a motor, it may not be enough to understand the difference in vibration before and after the failure. The clear differences that are identifiable the moment they are seen can be recognized even by maintenance technicians checking visually.

While monitoring vibrations, you can try to decompose their frequencies, or isolate the various elements in detail. In this way, how accurately can we establish the fact that a certain behavior is a sign of failure? The ability to analyze data is required. Of course, it is also essential to  refer to the knowledge and expertise of the maintenance technicians who have a wealth of actual hands-on knowledge.

In other words, IoT is hardware, but the software part, that is, how to analyze the data obtained from it and what kind of knowledge to establish, will also be extremely important. The implementation of IoT is not the end of the story, but rather the beginning of a more practical analysis that compares the collected data with the knowledge and insights gained in daily maintenance work. This is also connected to the suggestion that we should start small and implement the system as soon as possible to store as much data as possible.

--Lastly, how should we train maintenance technicians in terms of human resources as we move forward with the implementation of IoT?

Of course, it is also necessary to pass on the skills of maintenance technicians at the same time. It’s never about IoT or people or about choosing one over the other. The working-age population is going to decrease, and more jobs are becoming remote due to the new coronavirus. Maintenance work is no exception to this. The traditional way of passing on knowledge where senior maintenance technicians work together with younger technicians at the site may be replaced by remote instruction and education.

It was already a struggle for technicians to pass on tacit knowledge, but with remote instruction and education, it could take even longer than before to train maintenance technicians. There is a limit to how much you can teach remotely, and there are many detailed operations and nuances that are difficult to understand without being taught on site. This will make it more difficult than ever to pass on the skills of maintenance technicians, and there will be a need to devote resources to training them.

While it is important to collect and analyze data, the actual work of maintenance beyond that is left to human hands. Therefore, the training of maintenance technicians who can perform hands-on work on site will continue to be of high importance in the future.

This is where the process of identifying and formalizing each of the maintenance work tasks mentioned in Part 1 becomes important. This is a basic task that is the foundation for everything, both in implementing IoT and in training maintenance technicians. When considering the implementation of IoT, this will help you to think about which parts of the identified tasks should be included in IoT and which data should be captured to be effective. In human resource development, by verbalizing each task, the knowledge can be taught in a way that is easier to understand.

In order to succeed in both the implementation of IoT and the training of maintenance technicians, it is important to start with this kind of solid foundation.

Related articles

• Making Use of Sensor Fusion in Smart Factories
• Pitfalls in the Conversion of Regular Factories into Smart Factories Seen in Data
• Manufacturing DX from the Perspective of CPS Construction