How to measure tool life gains from a coolant change
Summary
Measuring tool life gains from a coolant change takes 5 steps: define tool life, fix conditions, set sample size, log other changes, and compare against variation. A gain counts only when the tool life criterion was fixed before the trial, both coolants ran on the same tools, parts and cutting data, and the difference exceeds edge-to-edge scatter.
ISO 3685:1993 tool life criteria for turning, as applied in one carbide study, end tool life at the first of several wear limits, including an average flank wear VBB of 0.3 mm and a maximum VBBmax of 0.6 mm, along with crater depth and notch wear limits (COBEM, 2009).
In that dry turning study, four upper edges of the same carbide inserts under constant cutting data gave mean tool lives from 977.8 to 1,298.5, a 33 percent spread, with no significant difference at 95 percent confidence (COBEM, 2009).
The standard deviation of tool life for a single edge position in that study ranged from 6 to 39 percent of the mean across eight edges, with three tests per edge (COBEM, 2009).
ISO 8688-1:1989 and ISO 8688-2:1989 cover tool life testing in face milling and end milling and were last confirmed in 2022 (ISO, 2022).
This method suits a production comparison of two metalworking fluids on one machining operation. It does not replace a laboratory tool life test to ISO 3685 or ISO 8688, and it does not cover grinding, where wheel wear is measured differently.
How do you measure tool life gains from a coolant change step by step?
Work through five steps, and fix each one in writing before the new coolant goes into the sump.
Define tool life for the operation: parts per edge until a stated wear limit or a stated part quality limit is reached. Use the same definition for both coolants.
Fix the conditions: tool grade and geometry, workpiece material and lot, cutting speed, feed, depth of cut, machine and coolant concentration.
Set the sample size: run a baseline on the current coolant, estimate the edge-to-edge scatter (how much tool life varies from one edge to the next), and decide how many edges each coolant needs before you compare anything.
Log other changes: record every change to tools, programs, material, machine, operators and coolant condition, with the date, for both periods.
Compare against variation: test whether the difference in mean parts per edge is larger than the scatter, using a two-sample comparison of means (NIST/SEMATECH e-Handbook, section 7.3.1).
Before the first edge runs on the new coolant, write down what result would count as a gain, for example a minimum percentage at a stated confidence level. A criterion or threshold chosen after the results are in turns the trial into a search for a favorable reading.
How should you define tool life when comparing two coolants?
Use one fixed end-of-life criterion, applied the same way to every edge on both coolants. That criterion is either a measured wear limit or a production limit, such as parts per edge until a dimension, surface finish or burr check fails.
Laboratory standards use wear limits. ISO 3685:1993 covers tool life testing with single-point turning tools, and one carbide turning study applied its recommended criteria as an average flank wear VBB of 0.3 mm or a maximum flank wear VBBmax of 0.6 mm, alongside crater and notch wear limits, whichever was reached first; maximum flank wear ended every test (COBEM, 2009). ISO 8688-1:1989 and ISO 8688-2:1989 cover the same task for face milling and end milling (ISO, 2022).
On a production line, measuring flank wear on every edge is often impractical, so many plants count parts per edge to a quality limit instead. That works if the limit is objective and the same for both coolants. The weak definition is "the operator changed the insert": operators change inserts for reasons other than wear, such as a shift change or a scheduled index. Record the reason for every tool change, and count only edges that reached the stated criterion.
Which conditions must stay fixed when measuring tool life after a coolant change?
Hold fixed the tool grade, geometry and supplier lot; the workpiece material and lot; the cutting speed, feed and depth of cut; the machine and program; and the coolant delivery, meaning nozzle position, pressure and concentration.
Cutting data matter most. Cutting speed has a strong effect on tool wear, so a small program edit can change tool life more than the fluid does.
Workpiece hardness varies between material lots and heats, so run both coolants on the same lot where you can, or record the lot for every part.
Treat coolant concentration as part of the condition, not a background detail. A water-miscible coolant that drifts lean or rich during the trial, below or above its set concentration, changes lubrication and cooling at the edge. Check and record concentration on a fixed schedule for both coolants, and correct it with the same rule. The choice between correcting a drifting sump and replacing it is covered in coolant concentration and pH drift: correct it or dump the sump.
If any of these conditions has to change during the trial, stop counting, record the change and start a new comparison period. Edges from before and after the change no longer belong to the same sample.
How long should a coolant tool life trial run, and how many tools does it need?
Run the trial until each coolant has reached the number of edges your baseline scatter calls for, not for a fixed number of weeks. That number depends on how much tool life varies from edge to edge, the size of gain you want to detect, and the risks you accept of a false gain and of a missed one (NIST/SEMATECH e-Handbook, section 7.2.2.2).
Tool life scatters widely even in controlled tests. In a 2009 dry turning study with constant cutting data, the standard deviation for a single edge position ranged from 6 to 39 percent of the mean across eight edges, from three tests per edge (COBEM, 2009). NIST notes that when the standard deviation is not known, the sample size calculation needs "perhaps the best estimate available from a previous experiment" (NIST/SEMATECH e-Handbook, section 7.2.2.2). On a production line, that previous experiment is a baseline period on the current coolant.
An illustrative calculation shows the scale. With the normal approximation, if edge-to-edge scatter is 20 percent of the mean and you want to detect a 20 percent gain, at a two-sided 5 percent risk of a false result and a 20 percent risk of missing it, the one-sample formula gives about 8 edges. A comparison of two independent groups needs about twice that, roughly 16 edges per coolant (illustrative). One edge per coolant, or a few days of running, cannot show a gain.
How do you rule out other changes when a coolant change seems to improve tool life?
Keep a dated log of every change on the operation for both periods, then check whether any logged change lines up with the improvement better than the coolant change does.
A production line does not stand still during a long trial. Typical candidates are a new tool lot or a different insert grade, a revised program or cutting data, a new workpiece material lot or supplier, machine maintenance such as a spindle or a new coolant nozzle, a cleaned and recharged sump, and changes in operators or shifts.
The changeover can be one of them. A full system clean before the new coolant fills the sump can itself improve chip evacuation and filtration, which may change tool wear apart from the fluid chemistry.
Three designs reduce the problem. Run both coolants at the same time on two comparable machines with the same tools and material, if the plant has them. Run the baseline immediately before the change, not months earlier. Or alternate periods, if the system allows it.
NIST describes a paired design, such as before and after measurements on the same objects, where each pair's difference is analyzed (NIST/SEMATECH e-Handbook, section 7.3.1.1). Laid out on matched machines or matched periods, pairing removes some of the between-condition variation.
What can you claim from a coolant tool life trial?
You can claim a gain in mean tool life, by the stated criterion, for the operation, tool, workpiece material and cutting data tested. The claim holds only if the difference is larger than the edge-to-edge scatter at the confidence level set before the trial.
Use the check table below to decide whether a tool life gain can be credited to the coolant change.
Check | Pass | Fail and what to do |
|---|---|---|
Tool life definition | One criterion fixed before the trial, the same for both coolants | Criterion set or changed after results: start a new comparison |
Fixed conditions | Same tool grade, workpiece lot, cutting data, machine and concentration log | A condition changed: split the data and compare only matched periods |
Edges per coolant | At least the number calculated from baseline scatter | Too few edges: keep running before claiming any gain |
Change log | No logged change lines up with the improvement | Another change coincides: treat the gain as unattributed |
Difference against scatter | Gain larger than scatter at the pre-set confidence (NIST/SEMATECH e-Handbook, section 7.3.1) | Within scatter: report no demonstrated difference |
The last row decides most trials: a percentage gain inside the scatter is not a result. The 2009 study above found a 33 percent spread in mean tool life between four edges with no significant difference (COBEM, 2009).
Published supplier cases often report a percentage without the criterion or sample size; one reports a 26 percent tool life gain after a changeover that began in December 2017 and took six months (Blaser Swisslube, case study, changeover from December 2017). Before you compare your result with such a figure, ask for the definition, edge count and conditions.
When should you call a machining specialist about a coolant tool life trial?
Call a machining or metalworking fluid specialist in three cases: the scatter in your baseline is so large that the required edge count is impractical; wear modes change between coolants; or the result will decide a supply contract or a plant-wide changeover.
A change in wear mode, the way edges wear out, is a strong signal. If the new coolant shifts edges from steady flank wear to chipping, built-up edge or notch wear, parts per edge may move for a reason that a single number hides. A specialist can examine worn edges, check whether the cutting data suit the new fluid, and propose a wear-based criterion for a short laboratory confirmation to ISO 3685:1993 or ISO 8688. A specialist also helps when the coolant change coincides with a tool or material change that cannot be undone.
Set the trial rules before the results arrive. Lubinpla's performance evaluation fixes the measurement criterion and sampling plan before results come in, and a criterion changed after results starts a new round. Whatever method you use, keep the criterion, the edge count, the change log and each edge's result in one record.
Frequently asked questions
Can you compare tool life records from a year before the change with the new coolant?
You can, but the comparison is weak: tools, material lots, programs and machines change over a year. Use the historical data to estimate scatter and set the edge count. Then run a fresh baseline on the current coolant just before the change, and compare with that.
Should you count tool changes from the tool management system as tool life?
Count them only if each change records its reason. A tool change made at shift end, by schedule or after a crash is not an end-of-life event. Filter the records to edges that reached the defined criterion; otherwise the data mix wear with routine changes.
Does a longer tool life prove the new coolant is better overall?
A longer tool life shows one benefit, on the operation tested. A coolant decision also depends on sump life, concentration stability, foaming, residue, skin compatibility and disposal. Report the tool life result with its limits, and judge those other factors on their own evidence.