Abrasive is not consumed hitting the part. It is consumed on the way back. In a closed machine, every grain passes through the circuit hundreds of times before it leaves — and what decides whether it leaves after a hundred passes or after three hundred is not the wheel. It is the separator.

It is the component nobody discusses at purchase. Wheel power, cabinet size, handling system — all covered. The separator appears as one line in the quotation, gets set once at commissioning, and is not touched again for years. Then the abrasive bill doubles and the blame goes somewhere else.

Where the abrasive actually goes

After impact, what falls to the floor of the cabinet is no longer the abrasive you put in. It is good abrasive, plus broken abrasive, plus oxide dust knocked off the part, plus mill scale, plus whatever else came off — burrs, foundry sand, sometimes pieces of metal.

That mixture is collected from the cabinet floor — screw, belt or hopper, depending on the machine — and lifted by a bucket elevator to the top of the unit. That is where the part that matters begins: cleaning. Only afterwards does the mixture reach the storage hopper and go back to the wheel or the nozzle.

If the cleaning stage is not doing its job, what returns to the wheel gets finer and dirtier with every cycle. Nothing stops, no lamp comes on. The cleaning cycle simply gets longer and consumption climbs.

The cascade separator: a curtain of abrasive and a stream of air

The principle is almost too simple. The mixture is allowed to fall freely, in a thin, even curtain, through a crosswise stream of air. Heavy, intact grains fall nearly vertically into the hopper. Broken grains, dust and scale flakes are light, so the air pushes them sideways and they end up in the filter.

Everything the separator does comes down to four things it controls: how the abrasive spreads, how it falls, how much air crosses it, and how much of the fines it manages to remove. Four things — and all four drift out of adjustment.

The curtain has to be thin. If the mixture falls in a heap, air never reaches the grains in the middle, and the broken grains in the middle of that heap go straight back into the circuit with the good ones. Likewise, if the curtain is uneven across its width, the thick zone is not cleaned at all, no matter how well everything else is set.

Too little air, too much air

Airflow through the separator has a window, not a vague optimum. Below it, fines stay in the circuit. Above it, the stream catches good abrasive as well and sends it to the filter — you are buying new abrasive to fill filter bags.

The second case is the more expensive one and the harder to catch, because nothing shows in the process: parts come out clean, cycle times look normal, only consumption is high. The evidence is in the filter dust. If you find whole grains in the dust bags rather than just powder, either the airflow is too high or the curtain is falling wrong.

The screen: what has no business in the circuit

The screen does not clean fines out of the abrasive — that is the separator’s job. The screen stops foreign bodies: burrs broken off parts, pieces of wire, clumped foundry sand from castings, forgotten bolts, lumps formed from abrasive and oil.

These do not change abrasive consumption, but they damage other things — they jam in blades, cause uneven wear, find their way into pumps. A blinded screen, on the other hand, sends everything into bypass, which is the same as not having one.

Magnetic separation: two opposite situations

A magnet separates by magnetism, not by size, and it is used in two scenarios that have nothing in common.

First: non-metallic abrasive — corundum, garnet — used on ferrous parts. The magnet pulls out the scale and swarf coming off the workpiece, which would otherwise build up and contaminate the abrasive.

Second, and the logic is reversed: stainless steel or aluminium parts, where a single trace of iron left on the surface starts rust within days. There the magnet is a safeguard on the circuit, not a saving.

The operating mix: the number that tells the truth

A healthy circuit does not hold abrasive of one single size. It holds an operating mix, with grains ranging from new to nearly spent, in proportions that stabilise after a few days of running. That mix, not the abrasive in the bag, is what works on the part.

A mix that is too coarse hits hard and produces a deep profile — and a deep profile eats paint. A mix that is too fine gives good coverage but a shallow profile and poor adhesion. The balance between them cleans fastest and gives the profile you specified.

Checking it means a sieve test: take a sample from the hopper, run it through a set of screens, look at the distribution. If the fine fractions grow month on month, the separator is losing ground. It is the only measurement in the whole process that tells you what is happening before it shows up on an invoice.

Signs the separator has stopped working

Cleaning cycles get gradually longer with no change in the parts.
Abrasive consumption rises with nothing in the process having changed.
Filter dust contains whole grains, not just powder.
Sieve tests show fine fractions growing month on month.
Parts come out clean but with a shallower profile than before.
The filter blinds more often than it used to.

Where the separator sits in total cost

Worth putting in context. In a typical breakdown of blasting cost, abrasive is the smallest line:

Item	Share
Labour	33%
Maintenance	22%
Depreciation	20%
Abrasive	13%
Energy	12%

The usual conclusion drawn from this table is that abrasive does not matter. That is wrong. Cheap abrasive affects 13% of the cost. A badly adjusted recovery circuit affects abrasive, labour and maintenance at once — because a degraded mix lengthens cycles, and the fines left in circulation wear blades, liners and ductwork.

What you can check in an hour, on the machine you already have

Stop and open the separator. Look at the abrasive curtain: is it falling thin and even across the full width?
Check whether the feed opening is worn or partly blocked — that is where an uneven curtain comes from.
Take a dust sample from the filter bags. If you find whole grains, turn the airflow down.
Take a sample from the hopper and run a sieve test. Keep it. Repeat in a month and compare.
Check the screen: blinded, torn, or running in bypass?
If you have a magnetic separator, check that it has been cleaned — a loaded magnet separates nothing.

None of it costs anything beyond an hour of downtime. Together they tell you whether the abrasive you buy every month is going onto parts or into the filter.

Abrasive recovery: separator, screen, cyclone