Impact Mill vs Air Classifier Mill: Which One Fits Your Material and Budget?

They look alike on a quotation and differ by one part. That part decides whether your D97 is a promise or an average.

By the EPIC Powder application team, reviewed by the grinding test lab · 3 November 2026 · 11 min read

Two machines, one purchase order, and a decision that stays in the plant for a decade. Start with the cut point, then the feed, then the budget, in that order. A 45 μm top size at 5 t/h is a plain impact mill problem. A 20 μm top size at the same tonnage is an air classifier mill problem.

The price gap between the two is real, and it is smaller than the cost of choosing wrong: a machine that cannot hold your D97 gets replaced, and a machine that over-grinds for ten years takes the difference out of your electricity bill. The numbers below let you run that comparison on your own duty.

01Let the cut point pick the machine

Your top-size tolerance picks the machine, and everything else negotiates the price. Three branches cover most of the inquiries that reach us, and each one has a clear loser as well as a winner.

100 μm and above: plain impact mill

At this band a screen or a hammer set does the job, and a wheel adds cost without adding capability. Hammer and pin mills hold 20–150 μm on mineral fillers in our own project experience, with throughput per kilowatt at its highest in the whole mechanical range. The same duty on a screen mill quotes 25–40% below an equivalent ACM.

20–45 μm: air classifier mill, sized for the top

This is the band where a screen stops being a specification. Screen aperture is a hole; the particle that leaves is a shape, and elongated particles pass a hole narrower than their long axis. A wheel sorts by drag against mass, so the cut follows the particle’s behaviour in air, whatever the hole size.

Below 20 μm: ACM at its edge, jet mill beyond

Mechanical classification stays economic down to about 5 μm. Below that the wheel has to run so fast that tip wear and heat become the limiting cost, and a jet mill takes over. Projects that sit on that line get a straight comparison before anyone signs.

02An impact mill is a class of machines, an ACM is one member of it

The two names sit at different levels of the same family tree, which is why quotations compare machines that were never equivalent.

An impact mill is a size-reduction machine that breaks particles by striking them with a fast-moving rotor, hammer or pin; the energy arrives as a collision, and the breakage follows from it.

That definition covers hardware from a hammer mill to a jet mill, and the entry on the subject divides milling devices into two general classes, impact and attrition, tracing the modern impact form to high-speed rotors, hammers and pins (Wikipedia, accessed November 2026; the underlying treatment is Swain, Patra and Roy, Mechanical Operations, Tata McGraw-Hill, 2011). The category describes the force. It says nothing about how the product size is controlled, and that is the part buyers actually need to know.

An air classifier mill is an impact mill with a rotating classifier wheel mounted in the same housing, so the cut point becomes a setpoint instead of a screen aperture.

Four variants you will be quoted

• Vasara mylly — hammers swing or are fixed against a breaker plate, with a screen at the discharge. The workhorse for 20–150 μm and the cheapest route to a coarse powder.
• Tappimylly — two pin discs, one stationary, and the material is caught between them. Narrow distributions on brittle feed, and a clean chamber that suits frequent product changes.
• Turbine or beater mill — a bladed rotor without a screen, used for coarse reduction ahead of a finer stage.
• Disc mill — toothed discs for tough and elastic materials that bounce off hammers instead of breaking.

Trade usage blurs the line. Suppliers write “impact mill” for any of the four, and “air classifying impact mill” for the version with a wheel. When a quotation says impact mill, ask whether a wheel is included, because that one question changes the machine, the achievable D97 and the price. If you want the full picture of the machine itself, we keep it on the air classifier mill specification walkthrough.

03The classifier wheel is the part you are actually paying for

Remove the wheel and you still have a grinder. What you lose is the top-size guarantee, and that is the only feature a specification cares about.

Size reduction continues for as long as particles stay in the chamber. A screen ends that stay by geometry: anything larger than the aperture goes back for another hit. A wheel ends it aerodynamically: particles above the cut are thrown outward by centrifugal force, particles below it pass with the air. Wheel speed sets the cut, so a change of product becomes a keypad entry, with no shutdown and no parts change. The direction of that relationship is predictable, since a faster wheel throws coarser particles outward and lowers the cut; the slope is machine-specific, and a test curve is a better guide than any formula.

D97 is the particle size below which 97% of the powder by volume falls, and it is the number a top-size specification is written against.

Specific energy is the electrical energy consumed per tonne of product, expressed in kWh/t, and it is the figure that separates a cheap machine from a cheap process.

Both numbers usually come from laser diffraction. ISO 13320:2020 fixes the procedure so that two laboratories differ by a known amount instead of an unknown one, which matters when your customer tests the incoming powder with a different instrument than the one you used to release it (ISO 13320:2020).

Without a wheel, residence time sets the size, and fines that already meet the spec stay in the chamber and keep being struck. The energy goes into narrowing a distribution nobody asked to narrow. With a wheel, those fines leave as soon as they qualify, and the mill spends its power on the coarse tail that still needs work. The mechanical difference between the two approaches is covered in more depth in pin mill and ACM particle-size control, side by side.

04Where a plain impact mill wins

Coarse, abrasive and frequently changing duties still belong to the plain impact mill, and on those duties the gap is wide enough to matter.

Coarse tonnage. At D97 100–300 μm no other mechanical mill moves more material per kilowatt. The rotor does not have to be precise, the screen does the sorting, and the whole assembly is built for volume.

Abrasive feed. Hammers and liners are simple castings. A set costs a fraction of a classifier wheel, and a worn hammer is swapped on a shift, and it never leaves the plant for rebalancing. On a Mohs 7 mineral, that difference in replacement cost shows up every quarter.

Fast changeover. A screen swap takes 10–20 minutes. Cleaning a classifier circuit and changing a wheel takes half a day, and on a plant running six products a week the accumulated hours are a real capacity loss.

Particle shape. Impact breakage produces angular particles with clean edges. For some fillers that raises packing fraction and is welcome; for powders that must flow through a 200 μm nozzle it is not. Shape is rarely on a datasheet and almost always visible downstream.

The limits are just as clear. There is no top-size guarantee, and there is always an oversize tail, because a screen passes the particle that fits and a real particle is not a sphere. On a 100 μm screen, a fraction above 100 μm gets through, and that fraction is what your customer’s sieve finds.

05Where the air classifier mill wins

At 10–45 μm with a contractual top size, the ACM is the cheaper machine across five years even though it costs more on the purchase order.

Capacity falls as the cut tightens, and the fall is steep enough to change a project’s economics. On the same frame, in our own project experience, moving a D97 target from 45 μm to 20 μm costs 35–50% of throughput and adds roughly 10–15 kWh/t. Ask for the capacity at your cut, with the test method named, and treat a brochure figure quoted at a coarser reference point as a different machine.

The practical ceiling is around 5 μm. Below it the wheel has to run fast enough that tip speed, heat and wear dominate the cost, and the economics move to a jet mill, which is when a jet mill earns its higher running cost.

06Four materials, four different answers

The material decides more than the price list does, and four common feeds split the decision cleanly.

Graphite: keep the flake intact

Anode graphite is judged on tap density and cycle performance, and both fall when flakes are rounded or broken. An ACM run at low rotor speed with the wheel doing the sizing keeps the platelet shape; a plain impact mill at high tip speed hammers it into rounds. For a 20 μm D97 with shape retention, the wheel is not optional. We wrote up the process in graphite anode grinding that keeps the flake intact.

Calcium carbonate: the forgiving one

Ground calcium carbonate at 10–45 μm is the easiest of the four. It is brittle, free-flowing and only moderately abrasive, so both machines work above 45 μm and the choice reduces to capital against controllability. Below 20 μm the wheel pays for itself within a year on most duties we quote.

Carbon black: feed is the bottleneck

Conveying is the hard part and grinding is the easy one: an over-fed chamber chokes before it does any work. A long-gap rotor with a wheel handles the low density better than a screen mill, which blinds as the material packs against the mesh. Soft feed also over-grinds quickly without a wheel, because nothing leaves until the air decides.

Bentonite: moisture decides

Above roughly 2% moisture the wheel blinds and the cut drifts coarse. Drying, or a heated air stream through the mill, is part of the machine choice, and it belongs in the budget. A screen mill blinds faster and is harder to clean, which is why bentonite tends to end up on an ACM once the deposit is variable. The bentonite line we commissioned in Greece needed inlet moisture controlled before the wheel would hold its cut for a full shift, and that control became part of the scope.

07What five years actually costs

The purchase order covers part of the cost. Electricity, wear parts and lost hours take the rest, and only one of the three appears on the quotation.

Split the number into five lines before comparing offers: the mill, the system around it (fan, filter, feeder, ducting, foundation), wear parts per year, energy per tonne, and downtime per changeover. The mill is the line buyers negotiate hardest and the line that matters least over a decade.

Energy is easy to under-estimate because the unit is small. Take 1,000 t/year at 30 kWh/t, which is 30,000 kWh. At US$0.10/kWh that is US$3,000 a year. Move to 45 kWh/t and it becomes US$4,500, a 50% increase that never shows up in a machine-to-machine comparison. Across five years the difference is US$7,500, which at this scale is often larger than the price gap between the two machines.

Wear follows the material, and the calendar has little to do with it. In our own project experience, abrasive minerals take a set of hammers in a few months, while soft carbon black runs for years on the same parts. Ask for a wear cost per tonne on your material, and treat a supplier who quotes only a parts price as incomplete.

Downtime is the line nobody puts in the spreadsheet. Two hours per changeover, twice a week, is 200 hours a year; on a line worth US$400 an hour in contribution, that is US$80,000, more than the difference between the machines in most projects we see. The MJW range and its working widths is where those changeover hours get priced.

08Twelve checks before you sign

Every one of these has killed a project we were later asked to rescue, so run them before the purchase order.

1. D97 and D50 both specified, with the test method named. A single number hides the distribution.
2. Capacity quoted at your cut, not at a reference cut. Ask for the derating curve.
3. Feed size upper limit in mm, and what happens above it. Pre-crushing may be cheaper than a bigger mill.
4. Moisture at the inlet, worst month rather than annual average. 2% is the usual wheel limit.
5. Hardness and silica content. Mohs 6–7 is where wear cost changes order of magnitude.
6. Bulk density and flowability, with a photo of the feed. Feeding problems look like grinding problems.
7. Specific energy in kWh/t at your duty. Convert it to annual cost before comparing prices.
8. Wear parts list with life in months and price per set. Then compute cost per tonne.
9. Changeover time and cleaning procedure, in hours. Multiply by your product mix.
10. Air volume and filter area, with the dust load. The system is quoted separately for a reason.
11. Turn-down ratio: lowest stable feed rate. Most lines run below nameplate most weeks.
12. Witnessed test on your material, with the report attached to the order. Numbers in a contract beat numbers in a brochure.

09What to send a supplier to get a real answer

Five items turn a quotation into a commitment, and a supplier who does not ask for them is quoting a catalogue.

Send the material name with an SDS, 5–10 kg of representative sample (not a dried lab scoop), the target D97 and D50 with the instrument method, the required kg/h and operating hours, and the feed conditions: moisture, bulk density, and the largest particle in the stream. Add what happens downstream, because a powder that flows badly through a bagging line fails in the plant even when it passes the sieve.

Our lab runs the sample through a pilot frame, measures the curve on a laser diffraction unit, and reports four numbers at your cut: achievable D97, throughput, specific energy and the frame that produced them. A curve without those four numbers is a catalogue figure with a test report stapled to it. Without it, every supplier is guessing at the same duty, and the cheapest guess usually wins the order.

Usein kysytyt kysymyksetQuestions buyers ask us