Statistical process control has a reputation as a large-shop, quality-department tool — the kind of thing that gets implemented because a Tier 1 customer requires it and then runs on autopilot without anyone understanding what it is actually doing. That reputation is mostly unfair. The core ideas behind SPC are practical and applicable at any scale, and applied to drill life management they solve a real problem: knowing when to change or resharpen a tool before it starts making bad parts, without changing it so early that you waste tooling.
What SPC Is Actually Doing
SPC is a method for distinguishing between two types of process variation. Common cause variation is the normal, expected scatter in a stable process — hole diameters that cluster around nominal within a predictable spread. Special cause variation is something different: a shift, a trend, or a spike that indicates the process has changed in some way that requires investigation or action.
For drill life monitoring, the special cause you are watching for is tooling wear. A worn drill shifts the hole diameter distribution, changes surface finish, increases burr size, or causes position drift. These changes are gradual but directional — they trend in one direction rather than scattering randomly. SPC gives you a formal way to detect that trend before it crosses your tolerance boundary.
The practical value is that it replaces two bad alternatives: waiting until you see a bad part (too late, you already made rejects), or changing tools on a fixed time schedule regardless of actual condition (wasteful if the tool had more life, and you are still guessing at the schedule).
Picking a Control Metric for Drill Condition
You need something measurable that responds to drill wear before hole quality goes out of spec. Hole diameter is the most direct choice for most applications — it is easy to measure with the gauging you already have, and it trends predictably as a drill wears. Other options depending on your application:
- Hole position: measured with a CMM or a simple go/no-go gauge relative to a datum. Useful when position tolerance is tighter than diameter tolerance.
- Burr height: harder to measure consistently but useful as a leading indicator when exit burr size matters for the next operation.
- Spindle load: if your CNC controller logs spindle current or load percentage, this is a direct proxy for cutting force and correlates with drill wear without requiring a measurement step. Some shops use this exclusively.
- Surface finish Ra: requires a contact profilometer or a calibrated visual standard. More useful in close-tolerance bore work.
Choose one primary metric. Hole diameter with a digital caliper or a bore gauge is the most accessible starting point for a small shop not running a formal quality system. Measure every nth part — every 5 parts in a fast run, every 10 in a slower one — and record the value.
Xbar/R Charts: Simplified
An Xbar/R chart is two charts stacked: the top tracks the average of a small sample, the bottom tracks the range (max minus min) within each sample. For drill monitoring, you do not always need true subgroup sampling — if you are measuring one hole per part, a simple individuals and moving range (I/MR) chart is more appropriate and simpler to calculate.
The individuals chart (I chart) plots each measurement directly. The moving range chart plots the absolute difference between consecutive measurements. Control limits on each chart are calculated from the data itself — you need 20 to 25 initial measurements from a stable process (sharp tool, normal conditions) to establish baseline limits.
The formulas without getting deep into the statistics:
- Center line (CL) = average of all initial measurements
- Average moving range (MR-bar) = average of the moving ranges from initial data
- Upper control limit for individuals (UCL) = CL + (2.66 × MR-bar)
- Lower control limit (LCL) = CL − (2.66 × MR-bar)
- UCL for moving range = 3.267 × MR-bar
These limits represent the expected variation of a stable, sharp-drill process. When ongoing measurements trend toward or exceed these limits, the process has changed — and the most common reason for a slow upward trend in hole diameter is drill wear.
Spreadsheet this. Fifteen minutes of Excel setup gives you a live control chart for any job you run regularly. Plot measurement number on the x-axis, dimension on the y-axis, and add three horizontal reference lines: CL, UCL, LCL. Look for trends — seven consecutive points trending in one direction is a classic SPC signal even before a control limit is crossed.
Setting Resharpen Trigger Points
The control limits tell you when the process has statistically changed. Your resharpen trigger point should be set inside those limits — when the trend indicates you are heading toward them, not after you have crossed them.
A practical approach: set a resharpen action limit at the midpoint between your center line and the control limit in the direction of wear (usually the UCL for diameter growth). When measurements cross this action limit or show a clear trending pattern toward it, the drill comes out for resharpening. This gives you early warning with enough lead time to act before bad parts.
For tighter tolerances, set the action limit closer to the center line. For loose tolerances, you can allow more drift. The chart makes this decision explicit and repeatable instead of relying on whoever is running the machine that day to make a subjective call.
Practical Drill Life Data Collection
The data collection burden needs to be low or it will not happen. The goal is a measurement that takes under a minute per sample, recorded in a way that does not require extra administrative work.
Options in order of simplicity:
Paper log at the machine. A printed sheet with part number, hole spec, and a column for each measurement. Operators record one value every N parts. At end of run, the sheet gets filed. Someone enters data into a spreadsheet later or it gets scanned. Low tech, but it works and it captures data even when digital infrastructure is not available at the machine.
Spreadsheet on a tablet at the machine. Operator enters measurements directly. Chart updates in real time. Slightly higher setup cost, lower transcription burden. A shared Google Sheet that any machine can access is adequate for this.
Machine data logging. If the CNC controller can log spindle load or cutting force data to a file, that data can be processed for SPC without any manual measurement step. This is more involved to set up but the data is continuous rather than sampled.
The data you actually collect consistently is worth more than a more sophisticated system you only use sometimes. Start with paper if that is what gets used. Upgrade when the process is established.
Closing the Loop: When to Resharpen, When to Retire
SPC tells you when to act. The action is either resharpening or retirement depending on what inspection of the removed bit shows. If the geometry is restorable — lip length adequate, no chipping, web not overthinned — resharpening makes sense. If the bit has been ground short from multiple prior sharpenings or has physical damage, replacement is the call.
A resharpened bit re-enters the rotation and goes through the same monitoring process. When it comes back from the resharpener — whether done in-house or through a mail-in service like MachinistPost — verify geometry before it goes back into production. Run five to ten holes and measure. Does the process return to the center line? If yes, the resharpen was successful. If measurements start off already shifted, inspect the grind. Inconsistent resharpening is a process variable just like wear, and SPC will catch it.
The shops that run this way — tracking hole data, acting on trends, cycling drills through consistent resharpening — consistently outperform shops running on gut feel and reactive tool changes. The math is not complicated and the data collection burden is low. The only requirement is doing it every run instead of just when something looks wrong.