Key takeaways
Store rooms label elements by a single number: 3 micron, 10 micron, 25 micron. The label implies a sieve, and the sieve model is wrong. A hydraulic element is a depth medium, glass or cellulose fibres millimetres thick, catching particles by interception, impaction and diffusion. Capture is probabilistic, rising with particle size rather than switching on at a threshold.
The beta ratio is defined at a stated particle size:
If a rig counts 20,000 particles larger than 5 micrometres per millilitre upstream and 100 downstream, beta_5 is 200. Convert with:
For beta 200 that is (1 minus 0.005) times 100, or 99.5 percent. Read the leakage rather than the efficiency: at the same particle size, beta 10 passes 10 percent of what arrives at the element and beta 1000 only 0.1 percent, a hundredfold cut per pass in what leaves the filter, not a hundredfold cut in the tank.
Beta is meaningless without its size. Beta 200 at 5 micrometres, written beta_5 = 200, is complete; "beta 200" alone is not, since the same element might be beta_12 = 200 and beta_5 = 2. Compare elements at the same size, chosen for the tightest clearance you protect.
Beta ratios come from a multi-pass test: contaminant is injected at a controlled rate while the element circulates test fluid, and counters sample upstream and downstream as differential pressure climbs to a terminal value. ISO 16889 fixes the method, the reference dust (ISO 12103-1 A3 medium) and the counter calibration (ISO 11171), but the test flow rate and the terminal differential pressure are chosen by the manufacturer and reported with the result, so two elements are comparable only when those reported conditions match.
The predecessor test, ISO 4572, used a different counter calibration and reference dust, so "3 micrometres" then and now are not the same particle population. Treat any figure with no stated standard as unverified.
Beta is a laboratory figure, measured on clean, steady flow: cold starts, pump ripple and cylinder surges leave field cleanliness worse than the bench predicts, so allow margin.
Filter selection is the last step. Identify the most contamination sensitive component, set a target ISO 4406 code, measure the oil now, then choose a rating and position that can hold it against real ingression. The three range numbers in an ISO 4406 oil cleanliness code map to counts at 4, 6 and 14 micrometres.
Two failures dominate. The first is matching the filter to the port size rather than the duty: a housing that fits the line says nothing about flow, viscosity or ingression. The second is skipping measurement, and repeatable oil sampling practice is what turns a target code into a control loop.
Dirt holding capacity is the mass of contaminant an element retains before terminal differential pressure. High efficiency with low capacity is a bad pairing: a beta 1000 element with small media area blinds within weeks in a dirty system, and the plant then fits something coarser.
Loading is signalled by a differential pressure indicator, and many assemblies carry a bypass valve cracking at 3 to 5 bar differential. Here is the trap that catches experienced people: once a filter is bypassing, the differential pressure across it stops rising, so the gauge looks stable while the element filters nothing. A steady, elevated reading is a prompt to check the bypass, not proof of health.
Change elements on differential pressure or on condition, never on the calendar alone, which discards good elements in a clean system and hides a bypassing one in a dirty system. The trigger should come from evidence, exactly as for bearing relubrication intervals. Record the ISO 4406 code before and after every change: if it does not move, the element was not the problem.
Ingression control is the other half. Dirt arrives through breathers, worn rod seals and sloppy top up practice, so fit desiccant breathers, replace weeping seals and use filtered transfer equipment. Structured lubrication routes stop those disciplines slipping. Filter changes and sampling belong in the same preventive maintenance plan as other routine work.
In Fabrico, a filter differential check can sit as a recurring PM task with a checklist, and each reading is stored against the asset, so a rising trend is visible. A failed manual check can trigger a follow up task, and machine stop data from a PLC or OEE feed can become a work order. To see per asset filter history, book a short demo.
An injection moulding cell runs a pump at 120 litres per minute into proportional valves. The valve maker asks for ISO 4406 18/16/13; the sample reads 21/19/15, three range codes worse at 4 and 6 micrometres and two worse at 14. The installed filter is a return line element rated beta_12 = 75, that is beta 75 at 12 micrometres, with no offline loop and a mesh breather cap.
Beta 75 is (1 minus 1 divided by 75) times 100, or 98.7 percent, at 12 micrometres. But the two code numbers that bind here are set at 4 and 6 micrometres, and the ISO 16889 curve gives only beta_6 = 2 there, or 50 percent: half that population passes on every transit. The element is competent near the 14 micrometre number and weak at the 4 and 6 micrometre numbers this system is actually failing.
Now the counts. Code 19 at 6 micrometres means 2,500 to 5,000 per millilitre, geometric midpoint the square root of 2,500 times 5,000, about 3,540. Code 16 means 320 to 640, midpoint 453. The required reduction is 3,540 divided by 453, roughly 7.8 times, and each halving is one range code.
Under a simple recirculating balance, the steady state concentration is proportional to ingression rate divided by (filtered flow times efficiency). Holding ingression and flow fixed, replacing the 50 percent element with one rated beta_6 = 200, that is 99.5 percent, raises the removal term from 0.5 to 0.995 and so cuts the steady state count by a factor of 1.99, about one ISO range code. The hundredfold figure belongs to the passing fraction, not to the oil in the tank, and confusing the two is the most common error in filter sizing: efficiency alone can never buy more than 2 times here, because you cannot remove more than everything that reaches the element. The 7.8 times the code target needs therefore has to come from more filtered flow and less ingression, which is why the fix below adds an offline loop and a desiccant breather rather than only a better element.
Position matters as much as rating, since the return line sees only what the actuators send back. An offline loop at 20 litres per minute on a 400 litre reservoir turns the tank over every 20 minutes, 3 turnovers per hour, where a beta_6 = 200 element of 90 grams capacity lives for months.
The verdict: keep the beta_12 = 75 return element as a coarse stage protecting capacity, add an offline loop with a beta_6 = 200 element, and fit a desiccant breather to cut the ingression that made the target unreachable. Re-sample after 50 running hours. The hardware change alone lifts the removal term from 120 litres per minute at 50 percent to that plus 20 litres per minute at 99.5 percent, about 1.3 times more removal, which is well under one range code. The rest of the improvement has to come from the breather cutting ingression, so expect a result in the 18/16/13 to 20/18/14 band, where the clean end needs the breather to cut ingression by roughly 6 times and the dirty end assumes it cuts it by only about 1.5 times, and read anything at the clean end as proof that ingression, not filtration, was the binding constraint.
Not on its own. A higher beta at the size you care about cuts leakage sharply, but it costs media area, pressure drop and element life. A beta 1000 element with low capacity blinds fast in a dirty system and gets replaced by something coarser, a net loss.
Work from the component you are protecting: choose a size at or below the tightest dynamic clearance, since particles near that dimension do the most abrasive damage. Proportional and servo valves usually push specification to 4 or 6 micrometres, and the target code should use the same sizes.
No. A micron rating describes an element; a code describes the fluid at a sampling point. What you achieve depends on the beta ratio at the relevant sizes, the filter position, the turnover rate, ingression through breathers and seals, and the wear debris generated.
A differential reading that has stopped climbing and now sits at or near the bypass crack pressure is the clue. Compare it with the bypass setting on the assembly drawing, and read it at operating temperature so cold viscosity does not mislead you. A downstream sample settles it.
It complements it. An offline loop holds bulk reservoir cleanliness efficiently, because flow is low, steady and unconstrained by the main circuit, so a high rating and a large element are practical. It cannot protect a component from debris generated upstream of it on a single pass.
Login
Blog