Spare Parts Optimization That Actually Works

The MRO Inventory Problem Nobody Wants to Talk About

Walk into any manufacturing plant's maintenance storeroom and you'll find the same thing: shelves of parts that haven't moved in three years sitting next to empty bins where critical bearings should be. The average manufacturer carries $500,000 to $5 million in MRO (maintenance, repair, and operations) inventory, and most reliability studies put the excess stock between 20-35% of that total. That's $100K to $1.75M in capital sitting on shelves, aging, and occasionally becoming obsolete when the equipment it was bought for gets decommissioned.

At the same time, stock-outs on critical spares cause 8-15% of all maintenance delays. A $40 bearing that's out of stock turns a 2-hour planned repair into a 3-day emergency because the part has to be expedited from a distributor 500 miles away at 3x the normal price plus overnight freight. The technician waits. The machine waits. Production waits. And someone writes a purchase order for 10 of those bearings 'so this never happens again,' which is how the overstock cycle perpetuates.

The fundamental problem is that traditional spare parts management is driven by fear and memory. A catastrophic stock-out event five years ago leads to permanent overstock on that part. A technician's preference for a specific brand leads to two equivalent bearings from different manufacturers both stocked, neither at the right quantity. Min/max levels get set during commissioning and never revisited even after failure patterns change. Nobody wants to be the person who reduced stock on a part that then causes a production-stopping breakdown.

Understanding Your Actual Demand Patterns

Before you can optimize spare parts, you need to understand why parts get consumed. MRO demand falls into distinct patterns, and each pattern requires a different stocking strategy. Treating all spare parts the same - which is what most min/max systems do - guarantees you'll overstock slow-movers and understock items with lumpy demand.

The category that causes the most pain is lumpy/intermittent demand - parts used only when something fails, with irregular intervals between uses. Standard min/max logic assumes relatively stable demand, so it either overstocks (min set too high based on a cluster of failures that won't repeat) or understocks (min set to 1, but when you need it, you need it now and the lead time is 6 weeks). This is where condition-based demand forecasting makes a measurable difference.

How Predictive Maintenance Changes the Parts Equation

The biggest shift that predictive maintenance brings to spare parts isn't a fancier reorder algorithm - it's advance notice. When a vibration sensor detects early-stage bearing degradation 4-8 weeks before failure, that advance notice completely changes your parts procurement options. Instead of carrying safety stock to cover unpredictable failures, you can order parts based on actual equipment condition and receive them before the repair is needed.

Consider a concrete example. You have 40 centrifugal pumps across your plant, each with two 6310-2RS bearings ($85 each from your distributor, 5-day standard lead time, $450 each expedited overnight). Under traditional stocking, you might carry 12 bearings in stock (min 4, max 12) because you replace an average of 8 per year but the timing is unpredictable - some quarters you use 4, some quarters you use 0. That's $1,020 in average inventory carrying cost, and you still get caught 2-3 times a year with an expedited order when demand clusters.

With condition monitoring on those 40 pumps, you get 4-8 weeks warning before a bearing needs replacement. Now you can carry a min of 2 (emergency buffer) and order against actual predicted demand. Your average inventory drops from 8 units ($680) to 3 units ($255). You eliminate expedited orders entirely. And you never have a stock-out because you're ordering against known upcoming need, not forecasting against historical averages.

That's a $1,850-$2,300 savings per year on a single bearing type. Multiply that across 500-2,000 intermittent-demand SKUs in a typical plant storeroom, and you're looking at $150K-$500K in annual savings from reduced carrying costs and eliminated expediting fees. The exact number depends on your asset count, failure rates, and current stocking levels - but 20-35% reduction in total MRO inventory value is consistently achievable within 12-18 months.

Lead Time Analysis: The Number That Drives Everything

The right amount of safety stock for any part depends on exactly one thing: the gap between how much warning you have before needing the part and how long it takes to get the part. If your predictive system gives you 6 weeks of warning and the part ships in 5 days, you barely need any stock. If your warning time is 2 weeks and the part has an 8-week lead time, you need to stock it regardless of how good your predictions are.

Most plants track vendor-quoted lead times but not actual lead times. The vendor says 3-5 business days; the actual time from PO to receipt might be 8-12 days when you account for approval routing, purchasing queue time, vendor processing, shipping, and receiving inspection. Map your actual end-to-end procurement cycle for your top 100 critical spare parts. You'll probably find it's 1.5-3x longer than the vendor's quoted lead time.

Implementing Condition-Based Reorder Points

Traditional reorder points are static: when inventory drops to the min level, generate a purchase requisition. Condition-based reorder points are dynamic: they increase when predictive signals indicate upcoming demand and decrease when equipment is running healthy. This is where the integration between your predictive maintenance platform and your CMMS inventory module (or standalone inventory system) does real work.

The implementation approach depends on your CMMS capabilities. In Maximo, you can use KPI-based reorder points tied to condition monitoring data - when a bearing health score drops below a threshold, Maximo can automatically adjust the reorder point for that bearing's part number. In SAP PM, you'd use MRP (Materials Requirements Planning) signals driven by maintenance order components - when a predicted maintenance order is created with a parts list, MRP picks it up in the next planning run. In simpler CMMS platforms like Fiix or eMaint, you'll typically implement this as a manual or semi-automated process where the reliability engineer reviews predicted upcoming repairs and generates purchase requisitions.

A practical consideration: condition-based reordering works best for parts used on your monitored, critical assets. For non-monitored assets - which will be the majority of your plant for years - you're still running traditional min/max or periodic review. Don't try to apply condition-based logic to parts that lack condition signals. A hybrid approach is normal and expected: condition-based for your top 100 critical parts, optimized min/max for the next 500, vendor-managed for consumables, and procure-to-order for insurance spares with acceptable lead times.

Tackling Dead Stock and Slow Movers

Every storeroom has a graveyard section: parts for machines that were decommissioned years ago, specialty items bought for a one-time repair that never recurred, and the optimistic bulk purchase that saved 15% per unit but left you with a 7-year supply. Dead stock (no movement in 24+ months) typically represents 15-25% of MRO inventory value. It's the easiest target for immediate savings because these parts are already a sunk cost - the only question is how to recover some value from them.

For slow movers (parts with sporadic demand - say, 1-3 uses per year), the question is whether to stock or convert to procure-to-order. The math is straightforward: compare annual carrying cost (typically 25-35% of part value per year) against the cost of expediting when needed (price premium plus express shipping plus downtime during wait). For a $200 part used once per year with a 5-day lead time, carrying cost is $50-70/year. If expediting costs $150 per event and you can wait 5 days, procure-to-order wins. If that 5-day wait costs $10,000 in lost production, stocking one unit at $70/year is obvious.

Building the Business Case for Parts Optimization

Spare parts optimization doesn't require a massive upfront investment, which makes the business case relatively easy to build - if you have accurate data. The challenge is that most plants don't know their actual carrying cost rate, their true stock-out frequency, or how much they spend on expediting. You'll need to pull these numbers before you can make a credible case.

Start by calculating your total carrying cost. Most CFOs use 25-35% of inventory value as the annual carrying cost rate, which includes: cost of capital (what you'd earn if that money was invested elsewhere, typically 8-12%), warehouse space allocation (rent, utilities, racking - typically 3-5% of inventory value), insurance (1-2%), obsolescence and spoilage (5-10% for MRO parts, higher for items with shelf life), and inventory management labor (cycle counting, put-away, issuing - typically 3-5%). If your finance team hasn't calculated a carrying cost rate for MRO specifically, use 30% as a conservative estimate.

The implementation cost for a spare parts optimization project is primarily labor - a supply chain analyst or experienced storeroom manager working with the reliability team for 3-6 months. If you're implementing condition-based reordering tied to a predictive platform, add the integration cost from the CMMS integration work (covered in our companion article). Total implementation investment is typically $40-80K in labor and $10-25K in software configuration. At $562K in annual savings, that's a 1-2 month payback - one of the fastest ROI items in a reliability improvement program.

Common Mistakes and How to Avoid Them

Spare parts optimization looks straightforward on paper but trips up in execution because it touches procurement, maintenance, finance, and operations. Each group has different incentives, and the optimization that makes sense from an inventory perspective sometimes conflicts with what maintenance or production needs. Here are the mistakes that derail otherwise solid programs.

• Cutting safety stock before predictions are reliable. If your predictive platform has only been running for 3 months, its failure predictions are not accurate enough to replace safety stock. Wait until you have 6-12 months of validated predictions with documented accuracy above 75% before reducing buffer inventory on critical parts.
• Optimizing by part number without considering assemblies. A centrifugal pump repair isn't just a bearing - it's a bearing, a seal kit, a coupling insert, and sometimes an impeller. If you stock bearings perfectly but run out of seal kits, the pump still doesn't get repaired. Optimize at the repair-kit level, not the individual part level.
• Ignoring vendor consolidation opportunities. If you buy the same SKR 6310-2RS bearing from three different vendors, you're splitting volume and losing leverage. Consolidate to 1-2 preferred vendors per category and negotiate volume pricing, consignment, and guaranteed availability.
• Setting it and forgetting it. Reorder points that were correct 12 months ago may be wrong today if you've added or retired equipment, changed production volumes, or shifted maintenance strategies. Review stocking levels quarterly at minimum.
• Letting procurement drive the policy alone. Procurement optimizes for unit cost and vendor terms. Maintenance optimizes for availability and lead time. These goals conflict. The right answer requires both perspectives plus the reliability engineer's input on predicted demand. Create a cross-functional review cadence.
• Not tracking the right metric. Inventory turns is the standard supply chain metric, but for MRO it can mislead. A part with 0.5 turns per year might still be correctly stocked if it's a $20,000 insurance spare for a critical compressor. Track stock-out rate, expediting cost, and inventory carrying cost separately - they tell you different things.