Could you walk into a manufacturing facility and get an accurate read on the maturity to plan and execute planned outages with a single question? Maybe the question you would choose would be around the SMRP Best Practice 1.5 Total Maintenance Cost as a Percentage of Replacement of Asset Value (RAV) to correlate the department's overarching responsibilities of asset management and their impact on EBIT. Or maybe you are the type that would ask an intrusive question that gets into the detailed processes that measure the effectiveness of predictive maintenance over time. I would challenge you to consider measuring their maturing to how they answered the question, what was the requested duration of their last planned outage. Do you think that you could get a good grasp of their maturity from this single question? I believe so.
A study conducted by Bernd Beber and Alexandra Scacco explored the possibility of surfacing election fraud by simply looking at the number of electoral returns. In their paper, What the Numbers Say: A digit-based test for election fraud, they indicated that the last digit in an electoral statistic should be random. Therefore, their position was that each number should end in zero as many times as it ends in seven. Referencing psychological papers and research as evidence, they indicated that humans could unintentionally reference specific single-digit numbers to try and make them appear random. Therefore, their position was that if certain digits appeared more than others at the end of a number, it could increase the probability that the electoral results had been manually manipulated. For example, 15743 might look like a normal five-digit number, but when you see 75423 and 47233, you might start seeing a pattern that the numbers end in three a lot more than randomly.
Fair election procedures should produce returns where the last digits occur with equal frequency, but laboratory experiments have shown that individuals tend to disproportionately select particular numbers, even when they have incentives to properly randomize. - Bernd Beber, Alexandra Scacco
Regardless of the industry, I have routinely seen the length of planned outages divisible by 8 hours. This would then make the planned outage duration the likes of 8, 16, or 72. Historically, this could have roots within the United States to when the Adamson Act of 1916 established an eight-hour day for railroad workers to receive overtime pay. This became the first federal law that regulated workers' hours within private companies when the United States Supreme Court upheld the constitutionality of this Act in Wilson vs. New 243 U.S. 332 (1917). Arguably, this 8-hour pattern could also be associated with Congress passing the Fair Labor Standards Act in 1938, which required employers to pay overtime to all employees who worked more than 44 hours a week. A few years later, this was reduced to a 40-hour work week and written into law. 40 divisible by 8, yes, if you start looking this 8-hour pattern starts to show up everywhere including the length of your outages.
Evaluate your next planned maintenance outage process and see if this pattern exists within your organization. If you see the requested or allocated duration divisible by eight, you might be allowing the duration to indicate the bookends of the planned outage versus having the maturity to allow the work to build the outage's duration. You may also be the department that says maintenance gets to work only within 8-hour shifts by default or the department that rounds the length of work to eight hours routinely. You may be part of an organization that sees the 16 hours allocated to the maintenance department because "that's how we have always done it." Or you may be the department that references an obscured sentence from the original equipment manufacturer (OEM) that indicated that the asset requires “exactly” 320 hours per year. You then divvied up this OEM reference into 8-hour chunks over the year. And when you run out, well... you run out of time to do planned maintenance. As the Soup Nazi says, "No soup for you."
But from a maturity perspective, the organization should be driving the reliability from a zero-base spend perspective and requesting the necessary time to conduct the planned work based upon a backlog of identified work and asset strategies. As maturity grows, the department transitions the scheduled backlog, resources, priorities, and criticality to request an optimized duration for the planned outage. It is not the other way around, where hours are given to maintenance for them to fill up work to the brim. Instead, the maintenance department should be presenting an optimum plan, with a duration that isn't always divisible by eight.
Note that I indicated the word “optimum,” because I am not indicating that the maintenance and asset management planning department rules all others when evaluating risk. Risks such as supply chain disruptions, scheduling of labor, or the synchronization of planned work with adjacent work are critical for the organization to succeed. Instead, I am indicating that this is an optimum request in exchange for a returned predicted reliability. You may see the maintenance department inevitably given a duration that is divisible by eight hours, but the requested duration should be not divisible by eight the majority of the time if Beber and Scacco’s study applies to the requested planned outage duration.
Take on the challenge to investigate if you are within an organization that exaggerates or rounds the duration of planned maintenance by eight hours because it feels right. Evaluate if the planning process is manually manipulating the duration or if the maintenance department always finds 16 hours of work to complete in a bookended allocation of 16 hours. When you find the departments that plan this way, challenge them to transition to requesting a finite duration. What you will be forcing the department to do is plan better. And with better planning, you will be maturing the abilities of asset management.