From Predictive Maintenance to Predictive Spare Parts Planning

Throughout this series, we have followed the journey from detecting potential equipment failure to being ready to act on it:

• Part 1 – From Maintenance Schedules to AI: We looked at how predictive maintenance evolved from historical reliability methods and condition monitoring to AI-supported prediction.
• Part 2 – Predictive Maintenance Gives You 30 Days. Now What?: We explored what needs to happen after an early warning, from identifying the correct spare part to checking availability and current supplier lead times.
• Part 3 – How Does a Machine Know It’s About to Fail?: We examined how sensors, condition data and AI can detect changes in machine behaviour and turn them into useful maintenance information.
• Part 4 – Robot Dogs, Drones and Smart Sensors: We looked at how mobile inspection technologies are expanding the ways manufacturers can collect condition data across their assets.

All of these developments are helping maintenance teams understand what may require attention and when. But knowing that an intervention is likely only solves part of the problem. The right spare part also needs to be available when that intervention takes place.

That brings us to the next step: Can the information generated by predictive maintenance also make spare parts planning more predictive?

Spare Parts Planning Has Traditionally Looked Backwards

Spare parts are inherently difficult to plan because demand for many MRO components is intermittent. A part may remain untouched for years and then suddenly become essential when a critical asset requires maintenance.

Planning decisions therefore typically draw on information such as historical consumption, existing inventory, supplier lead times, criticality and predefined stocking parameters. These inputs remain important, but historical demand has an obvious limitation: it tells us what was needed in the past, not necessarily what the equipment is likely to need next.

Predictive maintenance introduces a different type of information. If condition monitoring indicates that a component is deteriorating and an intervention is becoming likely, the organisation potentially has information about future spare-parts demand before the part is actually consumed.

That creates an interesting possibility: Could maintenance information itself become an input into spare parts planning?

What If Maintenance Could Become a Demand Signal?

We touched on this idea in Part 2. Research by Sha Zhu, Willem van Jaarsveld and Rommert Dekker examined whether planned condition-based maintenance could be used as advance demand information for spare-parts inventory control.

Rather than forecasting spare-parts requirements solely from historical consumption, their approach incorporated information about upcoming maintenance activities into inventory decisions. Using data from two large maintenance organisations, the researchers showed that this additional information could improve inventory control compared with the benchmark approaches considered in their study.

The broader principle is particularly relevant for manufacturers investing in predictive maintenance. If Maintenance already has information suggesting that an intervention is likely to take place, Inventory Planning does not necessarily need to wait until the spare part has been consumed before that information becomes relevant.

This would not replace historical consumption as an input. Instead, it could complement it with something traditional forecasting often lacks: an indication of what may be needed next.

The same shift is also visible from the OEM perspective. In our recent expert series with Sanne Egholm Klit, Head of Parts Solutions & Warranty at BEUMER, she described a future in which machine builders use data from sensors and maintenance systems to anticipate spare-parts requirements rather than waiting for customers to report a failure. As she puts it: “We would like to deliver spare parts on site before the customer knows that they need them.” [Read the full article: The Data Future: Deliver the Spare Part Before the Failure.]

Whether viewed from the operator or OEM side, the underlying idea is similar: better information about what is likely to happen to equipment can give the spare-parts organisation an earlier indication of what may be needed next.

How a 30-Day Warning Changes the Inventory Decision

Let's return to the bearing we have followed throughout this series. Condition data indicates that an intervention will probably be required within approximately 30 days, and the corresponding spare part is currently in stock.

At first glance, there is no availability problem. But imagine that the same bearing is used in several other critical assets, only one unit remains in inventory and the current supplier lead time is 60 days. Historically, consumption may have been low enough for a stock level of one to appear perfectly reasonable.

The new condition information changes that assessment. The organisation now has evidence that its remaining bearing is likely to be consumed within the next month, while replacing that inventory could take twice as long.

Nothing about the physical component has changed, and its historical consumption is exactly the same as it was yesterday. What has changed is what the organisation knows about likely future demand.

That information can potentially support a better inventory decision.

Predictive Does Not Mean Certain

There is, however, an important distinction between anticipating demand and knowing exactly what will happen.

A component showing signs of deterioration may remain operational longer than initially expected. Further analysis may change the diagnosis, or Maintenance may decide that an immediate intervention is unnecessary. A predictive-maintenance alert therefore cannot simply be translated into a purchase order or an automatic increase in stock.

This is one of the reasons predictive spare parts planning is more complex than connecting two systems and sharing a forecast. It requires organisations to use better information about likely future maintenance while still accounting for uncertainty.

Research into condition-based maintenance increasingly considers this connection between prognosis and operational decisions, including spare-parts management. The objective is not to turn an uncertain prediction into a certainty, but to use the additional information to make a better-informed decision.

Lead Times Change the Picture Too

Condition information is also only one part of the planning equation. As we explored in Part 2, supplier lead times can fundamentally change whether a component should be held in stock.

Imagine that a component could historically be sourced reliably within 15 days. Holding additional inventory may therefore have offered little benefit. If the actual lead time has since increased to 60 days, while condition monitoring typically provides around 30 days of useful warning before an intervention, the risk profile is very different.

This matters because lead times are not necessarily as static as the values stored in an ERP system suggest. Supplier capacity, sourcing routes and market availability can change, while planning parameters may remain untouched for long periods.

A spare-parts planning process therefore needs to consider the relationship between reliability, criticality, expected demand, available inventory and the current time required to replenish that inventory. When those inputs change, the appropriate stocking decision may change with them.

Predictive maintenance is becoming increasingly dynamic. Spare parts planning needs to be able to respond to changing information too.

From Static Parameters to Dynamic Recommendations

This is the principle behind SPARROW.Plan.

SPARROW.Plan combines reliability-based methods with operational spare-parts data to determine which components should be stocked and in what quantities. Factors such as historical usage, criticality and supplier lead times can be incorporated into the calculation, allowing recommendations to be recalculated when the underlying conditions change.

This is important because a stocking recommendation should not remain unchanged simply because a min/max value was entered into an ERP system years ago. A longer supplier lead time, inventory available elsewhere in the organisation or a change in the operational importance of a component can all alter the appropriate stocking strategy.

Information from maintenance and asset-condition systems could add another dimension to this approach in the future. If an organisation has increasing confidence that a particular component will be required within a foreseeable period, that information could potentially become another input into inventory planning.

The objective is not to use predictive maintenance as a reason to increase stock across the board. Quite the opposite: better information should help manufacturers distinguish more precisely between inventory that protects operational availability and inventory that simply ties up working capital.

Prediction Becomes More Valuable When Systems Connect

This brings us back to the broader argument running through this series. Predictive maintenance creates information earlier, but its value depends on whether the rest of the organisation can use that information effectively.

A predicted intervention can inform maintenance scheduling while also prompting a check of the relevant material data, available inventory across sites and current sourcing options. Over time, that information could also contribute to decisions about whether existing stocking levels remain appropriate.

For many manufacturers, these activities still sit in separate processes and systems. Reliability teams monitor equipment condition, Maintenance plans interventions, Inventory manages stock and Procurement handles sourcing. Each function may be making sensible decisions based on the information available to it, while valuable signals generated elsewhere in the organisation remain disconnected.

Connecting these areas does not mean that every decision needs to become automated. It means allowing information generated in one part of the maintenance process to improve decisions elsewhere.

For spare parts planning, that creates an opportunity to connect two questions that have traditionally been addressed separately: What is the equipment likely to need, and what should we have available when it does?

From Predictive to Proactive

Predictive maintenance has already changed the way manufacturers can think about equipment reliability. Instead of relying solely on historical failure patterns and predefined maintenance intervals, teams can increasingly incorporate information about the current condition of an individual asset.

Spare parts planning can evolve in a similar direction. Historical consumption, reliability, criticality, stock levels and lead times will continue to matter, but better information about expected maintenance requirements can add another useful signal to the planning process.

Predictive spare parts planning therefore does not mean knowing exactly which component will fail on which day. It means using the information available today to make a better decision about what should be available tomorrow.

And that brings us to the final question in our series.

If condition monitoring can identify deterioration, maintenance systems can anticipate an intervention, the required component can be identified and inventory planning can prepare for that demand, how much of this process could eventually happen automatically?

Next in the series: The Autonomous Factory: Can a Machine Order Its Own Spare Part?
‍
Missed the other parts of the series?

• Part 1 – From Maintenance Schedules to AI: We looked at how predictive maintenance evolved from historical reliability methods and condition monitoring to AI-supported prediction.
• Part 2 – Predictive Maintenance Gives You 30 Days. Now What?: We explored what needs to happen after an early warning, from identifying the correct spare part to checking availability and current supplier lead times.
• Part 3 – How Does a Machine Know It’s About to Fail?: We examined how sensors, condition data and AI can detect changes in machine behaviour and turn them into useful maintenance information.
• Part 4 – Robot Dogs, Drones and Smart Sensors: We looked at how mobile inspection technologies are expanding the ways manufacturers can collect condition data across their assets.
• Part 6 – The Autonomous Factory: Can a Machine Order Its Own Spare Part? We bring the series together and ask how far the process from detecting deterioration to identifying, locating and procuring the right spare part could eventually be automated. (to be published soon).