CNC Coolant Cleaning Solution: How Filtration Actually Works

• 1
  What a CNC coolant cleaning solution actually has to remove
• 2
  The four mechanisms inside a coolant cleaning system
• 3
  Concentration, pH and why filtration alone is not enough
• 4
  When a CNC coolant cleaning solution pays back, and when it does not
• 5
  Coolant cleaning devices compared

Coolant leaves the drum clean. Then it does its job: it cools the cutting edge, carries chips away from the zone, and leaves an oily film on freshly cut metal. Every one of those functions adds something to the fluid. The job of a CNC coolant cleaning solution is to take those additions back out without stripping the fluid of the properties you paid for.

Four contaminants matter in practice. Tramp oil arrives from way lube, spindle oil and hydraulic leaks. Fines are the sub-20 μm particles that a chip conveyor never catches. Bacteria and fungi grow where the oil layer seals oxygen out. Dissolved salts and hardness minerals come in with the make-up water. Each one needs a different mechanism, and no single device handles all four equally well.

The failure mode is rarely sudden. A sump that held 12% concentration in January reads 6% in April because evaporation removes water and the refractometer reading drifts with contamination. Foam builds, smell turns sour, and the operator adds water to stop the foaming. That dilutes the fluid further. Two weeks later tools start chipping and nobody connects it to the tank.

So the honest question is not whether to filter. It is which fraction you need to remove first, because that decides the hardware. A shop fighting oil needs a skimmer. A shop fighting fines needs media filtration. A shop fighting bacteria needs aeration and concentration discipline before it buys anything.

Gravity separation is the cheapest stage. Free oil rises because its density is roughly 0.85–0.90 g/cm³ against 0.95–1.05 g/cm³ for water-based coolant. A quiescent tank with a 20–30 minute residence time lets a large part of the tramp oil float to a weir. This costs almost nothing to run and removes the bulk load before anything else sees it.

Coalescing plates accelerate that rise. The oil droplets hit inclined polypropylene plates, merge into larger drops, and climb faster. Plate spacing of 6–12 mm with a 45–60° incline is typical. Below about 30 μm droplet size the plates stop being useful and you are into chemistry or membranes. This is the boundary most shops hit and misread as filter failure.

Media filtration handles the solid fraction. Bag filters at 5–25 μm, cartridge filters at 1–10 μm, or a diatomaceous earth bed for finer cut. The trap is that a 5 μm bag on a dirty sump blinds in a shift, and every change costs labor plus the coolant carried out with the bag. Depth media buys run time; surface media buys clarity.

Centrifugal separation splits by density difference and handles both oil and fines down to roughly 2–5 μm, but the bowl needs cleaning and the flow rate is fixed by geometry. Membrane and reverse osmosis systems sit at the far end. They produce very clean water, but they also strip the additives, so you must reconstitute the concentrate afterward. That is a chemistry commitment, not a filter purchase.

Refractometer readings on emulsified coolant lie once contamination climbs. Tramp oil and dissolved salts shift the refractive index, so a true 8% emulsion can read 11%. The correction: titrate a sample or use a refractometer factor against a known dilution. Measure at the same time each week, from the same point in the tank, after the pump has run for 10 minutes.

pH is the leading indicator of trouble. Fresh emulsion sits near 9.0–9.5. When it drops below 8.5, the corrosion inhibitor package is depleting and odor usually follows within days. Do not fix this by dumping biocide. Find the oil layer first, because anaerobic bacteria live under it. Break the layer and the population collapses on its own.

Conductivity tells you about water quality. Hard make-up water leaves scale and consumes the emulsifier. If your conductivity climbs steadily between top-ups, switch the make-up to deionized or softened water. The cost is small next to a full sump change on a 4,000 mm machine.

Skim oil daily, aerate continuously, top up with premixed emulsion rather than water, and keep records. A shop that does those four things can run a sump 6–12 months. A shop that only bought a filter and skipped the discipline will still dump the tank twice a year.

Run the numbers on your own sump. Count sump volume, how many tanks, sump change frequency, the cost of a charge of concentrate, and the labor hours per change. Then add the tool cost. Dirty coolant shortens tool life and raises scrap on tight-tolerance work, and that line item usually dwarfs the fluid bill.

Small machines with short campaigns rarely justify a skid. If you run one 500 × 500 × 450 mm machining center two shifts a week, a portable tramp oil skimmer and a 25 μm bag housing cost little and capture most of the benefit. Spend the rest on aeration.

Large cells with continuous production are the opposite case. Six or more machines sharing a central sump, running 24/7 on aluminum or cast iron, generate tramp oil and fines faster than any manual routine can remove them. A coalescer plus a centrifuge or a DE bed pays back in reduced sump changes and steadier tool life.

Material matters too. Cast iron sheds graphite and fine swarf that loads media fast. Aluminum produces fine, light chips that float and resist settling. Titanium and Inconel generate heat that degrades the emulsion faster, so cooling and aeration become the priority over fine filtration. Match the mechanism to the swarf you actually make.