Simplify maintenance decisions

AI and decision fatigue in industrial maintenance

Maintenance decision fatigue poses a threat to safety, reliability and productivity. One way organizations can reduce decision fatigue is utilizing connected technologies to simplify the decision making process, so that maintenance can focus on the decisions that matter.

Decision fatigue, the diminishing quality of decisions a person makes after he or she has already been making a series of decisions for a long period of time, can substantially impact maintenance performance. In an ideal world, maintenance personnel would make one perfect decision after another about asset maintenance to improve availability and reliability throughout the plant.

In reality, decision fatigue can derail the best laid maintenance plans and strategies. It’s not simply that plants have little or no control over how maintenance workers are spending their time outside of work, or how many decisions they’ve had to make before they walk through the plant gates. Rather, it’s the fact that maintenance work intrinsically sets workers up to experience decision fatigue.

An abundance of work orders can create decision overwhelm

While routine aspects of maintenance can be delegated to less experienced maintenance professionals, more complex root cause analyses and repairs require the expertise and know-how of specialists. A junior maintenance worker will not be able to draw conclusions from process or vibration data analysis or ocular even observations, hypothesize the root cause or confirm a diagnosis. The complex matrix of decision making involved in troubleshooting and repairing an asset can trigger decision fatigue.

Production goals are driving faster but not necessarily better decision making

Unscheduled downtime forces maintenance into emergency mode, as engineers and technicians rush to troubleshoot the issue and reboot the asset to minimize production losses. This concentrated period of decision making, driven by pressure to reduce the unplanned downtime as quickly as possible, can lead to less efficient decisions being made during and after the issue has been resolved.

When overtime is required to solve the problem, decision fatigue can be compounded with physical fatigue, reducing maintenance personnel’s ability to think about the failure in a critical or creative way. As a 2018 National Center for Biotechnology Information article on mineworker fatigue argues:

If a worker is already in a depleted state (i.e., fatigued), there may not be a sufficient amount of mental resources for the worker to engage in self-regulatory actions and poor decision-making may result. This is somewhat connected to the Conservation of Resources Theory (Hobfoll, 1989), which argues that stress is the result of loss of physical or psychological resources, and that stress is itself resource depleting.

Maintenance decision fatigue is not sustainable

Decision fatigue can have serious consequences for major maintenance KPIs. Perhaps the most obvious example is safety – a maintenance technician’s decision fatigue can have consequences ranging from minor injury to lost time injuries or in the worst-case loss of life. Similarly, poor decisions dictated by decision fatigue towards the end of a shift, a night shift or an overtime shift can lead to less than adequate repairs and continued asset distress.

In the long-term, relying on senior maintenance technicians to make numerous critical decisions daily or weekly is not sustainable because these senior-level engineers are retiring. In the US, 2.6 million baby boomers are expected to retire by 2028, and it is unclear who will fill these roles and when.

Finally, it is worth noting that the next normal has led to fewer technicians on the plant floor at any given time. That means that in many cases fewer technicians are expected to keep production up and running, adding new decision-making responsibilities to technicians that may already have been at capacity.

Connected technologies can alleviate decision fatigue in process industries

Though decision fatigue may not be top of mind for most plants, manufacturers are very aware of fatigue and the dangers it poses to employees and assets. Safety risk assessments and procedures aim to prevent fatigue, either physical or mental, from causing employees to err and injure themselves or others. HR policies that are mindful of the latest research on night shifts and fatigue and work to minimize these shifts per worker can similarly improve fatigue and productivity.

Still, when it comes to decision fatigue, a plant cannot significantly reduce the number of decisions that need to be made regarding routine and emergency maintenance. What it can do is use connected technologies and Remote Maintenance to simplify decision making so that senior technicians are faced with fewer decisions – critical and non-critical – throughout their shifts.

Discover faults earlier to reduce overtime. Overtime is a standard side-effect of reactive maintenance. Condition monitoring for vibration analysis and Machine Learning-based analytics for process data reduce the need for overtime by letting technicians know days, weeks and sometimes months in advance that a failure is under way.

These early warnings enable maintenance to proactively order parts, plan the repair, and schedule specialists to remediate during daytime shifts. By minimizing physical fatigue triggers such as nighttime shifts and overtime, and reducing the pressure of resuming production associated with unscheduled downtime, maintenance personnel can consistently make better decisions that extend asset health and availability.

Time to failure estimations ease repair prioritizations. Triaging critical assets for repair can be complicated, especially if there is a shortage in asset specialists. Time to failure estimations provided by Machine Learning predictive maintenance can be the differentiating factor.

An asset with a time to failure estimation of 21 days can be attended to later than an asset with a failure warning of 14 days. Scheduling downtime according to failure criticality limits the number of important decisions a senior maintenance professional needs to make on a shift and contributes to better decision-making processes.

Consulting remote diagnostic centres simplifies decision making. Certain maintenance tasks, such as processing and analyzing vibration fault data, require concentrated periods of decision making by experienced engineers. Contracting a remote diagnostic centre to provide root cause analysis and corrective actions based on this data decreases the number of decisions these engineers need to make and enables them to focus on other aspects of the upcoming repair.

Decisions don’t have to reach maintenance in the first place. The early failure warnings and data insights provided by vibration analytics and Machine Learning predictive maintenance can enable operations engineers to adjust process variables to resolve the issue before it impacts product quality. In this case, the decision can be handled by operations with little or no intervention from maintenance, reducing maintenance’s decision overload.

Ring formation in cement rotary kilns can be counteracted and resolved with early detection. See how SKF Enlight AI’s AutoML predictive maintenance solution is helping cement plants foresee and prevent unplanned downtime of rotary kilns here.

Summary and conclusion

Throwing digitalization projects at decision fatigue and hoping that will solve the problem is wishful thinking. The move to combat maintenance decision overwhelm must be coordinated throughout the organization, and should include a mixture of process and safety reviews as well as training to recognize and prevent decision fatigue at the individual level.

However, connected technologies can augment these organizational initiatives by reducing the number of decisions maintenance personnel face and providing them with data-driven insights that simplify the decision-making process. As plants move to identify new ways to improve employee safety within this next normal, tackling maintenance decision fatigue and overwhelm is an important part of the conversation.

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.