When asked about optimising maintenance, different angles can be explored.
First, maintenance optimization can target the outcome; i.e. better maintenance. Better maintenance should result in less failures between maintenance intervals. Additionally, better maintenance can target lower costs for the same outcome. Business process optimization is the usual approach to achieving these targets.
Today, I’d like to highlight two aspects of maintenance optimization that are at the heart of predictive analytics projects for maintenance. The first, and typically the one corporate executives think of when starting predictive analytics initiatives, is to lower the frequency of maintenance.
This can be done by spacing out the intervals equally, more often than not through a relaxation of the safety levels. Mind you, this is not necessarily bad as in many cases, the maintenance interval is the result of safety upon safety upon safety which leads to overly conservative maintenance requirements. A better approach is to introduce condition based elements in order to allow for the delay of a maintenance event based upon the observed condition of the equipment.
This leads to certain intervals being more spaced out than others due to many factors such as: what is produced, how many, what materials,… Correct evaluation of the condition of the equipment is complex and cumbersome. Instead of relying on manual processes, sensors, diagnostics, and finally predictive systems now allow us to automate the process of assessing the condition and remaining cycle time before the next maintenance. A good system will warn with enough lead time to allow for planning the maintenance event.
All this is very nice but it only focuses on planned maintenance. The bulk of the pain/cost comes from unplanned events though. And that’s also where the biggest savings potential comes from. Gathering condition, usage and circumstantial data actually allows us to get pretty good visibility on future unplanned events. Avoiding them by planning a pre-emptive maintenance turns these into planned events with minimal impact to the production schedule. Given enough lead time, the pre-emptive maintenance can be made to coincide with a planned maintenance, lowering production impact even further.
While the ultimate goal is to space out maintenance events (higher availability for production and less maintenance cost), there are some hurdles to be overcome before getting there:
– resistance from the maintenance teams: “we’ve done it a certain way for so many years, why change?”
– resistance from management for taking responsibility
– legal hurdles: certain markets (i.e. passenger transport) are heavily regulated and the regulator needs to be convinced on the validity of the new approach
This is why (in most cases) the initial focus should be on eliminating unplanned events. Besides the lower barriers to get there (organisational, regulatory,…), the potential cost savings are also higher! Let’s make sure non-specialised executives are aware of this.