← Back to Field Notes

March 12, 2025 · The Bloomfield Team

How to Build a Preventive Maintenance Program from Scratch

Unplanned downtime costs the average discrete manufacturer between $10,000 and $250,000 per event, depending on the machine, the job it was running, and how long it takes to get it back online. Aberdeen Research puts the industry-wide average at $260,000 per hour for automotive and aerospace manufacturers. Even for a 30-person job shop, a single spindle failure on the wrong machine at the wrong time can cost $15,000 to $40,000 in lost production, rush repair charges, and missed delivery penalties.

A structured preventive maintenance program reduces unplanned downtime events by 25% to 40% in the first year. The math on that is straightforward. If your shop experienced 24 unplanned downtime events last year at an average cost of $8,000 per event, that is $192,000 in losses. Cutting 30% of those events saves $57,600. The PM program that produces that result costs far less to implement.

Step 1: Build the Equipment List

Start with a complete inventory of every piece of production equipment, including CNC machines, manual equipment, support systems (air compressors, coolant systems, chip conveyors), and material handling equipment. For each asset, record the manufacturer, model, serial number, install date, and current condition assessment.

This list becomes the backbone of the program. Every task, schedule, and cost record will tie back to a specific asset. Most shops have this information scattered across purchase orders, insurance records, and the maintenance person's notebook. Consolidate it into a single document or spreadsheet before doing anything else.

Step 2: Gather Manufacturer Recommendations

Every CNC machine, every compressor, every hydraulic unit comes with a maintenance manual that specifies service intervals. Haas recommends weekly spindle warmup verification, monthly coolant concentration checks, and annual way cover inspection. Mazak publishes specific lubrication intervals for every axis drive. DMG MORI provides maintenance schedules tied to spindle hours.

Collect these recommendations for every machine on the floor. They form the starting point for your PM schedule. You will adjust them based on your operating conditions (a machine running 20 hours a day in a high-chip-load environment needs more frequent service than the same machine running 8 hours on light aluminum work), but the manufacturer baseline gives you a defensible starting point.

Step 3: Prioritize by Criticality

Every machine on the floor is not equally important to production. Rank each asset by three factors: revenue impact if it goes down, likelihood of failure based on age and condition, and availability of backup capacity.

A single five-axis mill that runs 60% of the shop's aerospace work and has no backup is a top-priority asset. A manual Bridgeport used for deburring operations three times a week is lower priority. Allocate PM resources accordingly. The program does not need to cover every asset at the same level of detail on day one. Start with the five machines whose failure would hurt the most and build outward.

Step 4: Create the Task Library

For each priority machine, build a list of specific maintenance tasks with three attributes: what to do, how often to do it, and how long it takes.

• Daily: Check coolant levels, verify chip conveyor operation, inspect way covers for damage, check hydraulic pressure gauges (5 to 10 minutes per machine)
• Weekly: Clean and inspect tool changer mechanisms, check spindle warmup parameters, verify lubrication system levels and flow (15 to 30 minutes per machine)
• Monthly: Test coolant concentration and pH, inspect and clean electrical cabinet filters, check belt tension on spindle drives (30 to 60 minutes per machine)
• Quarterly: Full axis backlash check, hydraulic fluid analysis, spindle bearing temperature trending (2 to 4 hours per machine)
• Annual: Full geometric accuracy verification, ballbar testing, complete lubrication system flush and refill (4 to 8 hours per machine)

Keep the task descriptions specific enough that any qualified maintenance person can execute them without interpretation. "Check coolant" is not a task. "Measure coolant concentration with refractometer, target 6% to 8%, adjust with concentrate if below 5%" is a task.

Step 5: Schedule Without Wrecking Production

The biggest reason PM programs fail in small shops is that production always wins the scheduling fight. The machine is booked. The job is due Friday. The PM gets pushed to next week, then next month, then never.

Build PM time into the production schedule as a standing commitment. If a machine needs 30 minutes of weekly maintenance, that 30 minutes comes out of the available production hours for the week. The scheduler sees it. The operator knows it. The shop manager treats it as non-negotiable.

Some shops run PM during shift changes. Others dedicate Friday afternoons. The specific timing matters less than the consistency. A PM program that runs 80% of its scheduled tasks on time outperforms a PM program that runs 40% of its tasks whenever production allows.

Step 6: Track Results and Adjust

Measure two things from day one: PM task completion rate and unplanned downtime events. Both should move in the expected direction within 90 days. Completion rate should climb above 85% as the team builds the habit. Unplanned events should begin declining as the most common failure modes get addressed through regular inspection and service.

After six months, you will have enough data to refine the schedule. Some tasks will prove unnecessary at their current frequency, you can extend them. Others will reveal themselves as insufficient, a quarterly check on a component that fails every two months needs to become monthly. Let the data drive the adjustments.

For a broader look at how operational data connects to decision-making, see our guide to production visibility.

Common Mistakes to Avoid

Trying to implement a full PM program across every asset simultaneously overwhelms the maintenance team and produces poor compliance. Start with five to eight critical machines, prove the process works, and expand.

Assigning PM tasks to operators without training them produces checkbox compliance without actual maintenance value. An operator who checks "coolant OK" without using a refractometer is generating paperwork, not data. Train the team on the specific procedures and the reasons behind them.

Running a PM program on paper forms works for the first month and degrades after that. Digital tracking, even a basic shared spreadsheet with completion dates and notes, creates accountability and makes trend analysis possible.

The goal of a preventive maintenance program is to make unplanned downtime rare enough that it surprises people when it happens. That standard is achievable for any shop willing to commit the resources and discipline it requires. The machines already tell you what they need. The work is building the system that listens.