How to scale Maintenance 4.0

Scaling Predictive Maintenance: What Not to Do

Adopting and successfully implementing a Maintenance 4.0 program requires top-down and bottom-up buy-in. Leaving the Machine Learning technology development and deployment to vendors is a smart move, allowing these and other vendors to dictate your strategic vision is not.

As industrial producers seek to adopt Industry 4.0 Predictive Maintenance practices, there are several nice-to-haves that are not critical. In some cases, risk is associated with waiting until the entire technology roadmap is completed or until the expensive infrastructure is deployed. Maintenance 4.0 Predictive Maintenance requires both agility and discipline. These sometimes make for strange bedfellows.

Specifically, we do not believe that the following elements are necessary in a framework for scaling:

  1. Allowing Vendors to Frame the Strategic Vision: Because industrial plants lack experience with Industry 4.0, the temptation may arise to rely on large vendors’ expertise. This is a mistake. A strategy should be developed independent of outside influence, especially that of third parties whose agendas (and incentives) do not align with those of the plant.
  2. Paying Attention to the Market: With the accelerated rate of innovation in Machine Learning and Artificial Intelligence, it is easy to become overwhelmed by the number of new market entrants and solution offerings.
  3. Relying on Citizen Data Scientists: The term “Citizen Data Scientists” has been popularized by Gartner in the last few years. The idea is that, with sufficient training, semi-qualified technicians can become proficient enough in Big Data science to perform low-level tasks. In our opinion, given the increasing complexity of data science, the Citizen Data Scientist cannot fill the skills gap created by shortages of Big Data professionals.
  4. Building an Internal Machine Learning Discipline: In the short to medium term, very few industrial plants can build deep competencies in Machine Learning.  The field of Machine Learning is dynamic and fast-changing.  Although institutional knowledge about Machine Learning is relevant in an era of digital transformation, it should not be a dependency for change.
  5. Overemphasizing the “Digital Culture”: A Digital Culture is an aspirational concept that will take time to achieve.  The importance of a Digital Culture is recognized, but it is seen as an outcome of adopting Maintenance 4.0, not as a key enabling factor in the short term. 

There are also strategic mistakes you should be actively trying to prevent in a scaling framework:

  1. Failing to Evangelize: Change does not happen on its own and it is not uncommon for plant level Operations and Maintenance employees to be resistant to new technologies and processes. This is particularly the case if there is a perceived threat to employees’ job security. Scaling Predictive Maintenance solutions requires commitment and buy-in, which is why there is a need for internal champions / evangelists.
  2. Undisciplined Procurement Processes: Selecting vendors requires discipline. By failing to formalize their vendor management processes, organizations may end up using subjective criteria for vendor selection and over-emphasize “relationship” factors. Standardized processes for vendor selection and penalties for underperformance should be clearly communicated and consistently applied.

Poorly defined pilots: A recent survey by Cisco indicated that 60% of IoT initiatives do not go beyond the pilot phase. Why is this? One contributing factor is that it is a lot easier to agree to a pilot than to make the decision to move forward. Pilots should only be conducted when there is an agreement with all the internal stakeholders to move forward with deployment if specific pilot objectives are achieved.

Summary and Conclusion

It has been observed that “75 percent of plant performance improvement is attributable to human and managerial skills.” Technology cannot be scaled without plant level employee buy-in. Many of the factors preventing scaling of Machine Learning-based Predictive Maintenance can be addressed with careful planning and assigning the right people to spearhead change.

SKF Enlight AI

Industrial plants generate terabytes of process data. SKF Enlight AI is a SaaS Predictive Maintenance solution that uses Automated Machine Learning to identify emerging asset failure patterns within this data. It provides early warnings and sensor-level intelligence to help avert unplanned downtime and meet production goals. For more information on how SKF Enlight AI can improve performance and reliability, click here.

Leave a Reply

Your email address will not be published.