Fluid Cleanliness and Component Life in Closed-Circuit Hydrostatic Drives

2. The closed hydrostatic loop

2.1 Architecture

In a closed circuit the two working lines of the pump connect directly to the two ports of the motor. There is no reservoir in the main power path: the same fluid recirculates, and reversing the direction of pump displacement reverses the direction of motor rotation without any directional valving. What leaves the loop leaves it through leakage — past pistons, slippers and valve plates into the pump and motor cases — and through the loop flushing (hot-oil shuttle) valve, which deliberately bleeds a fraction of the low-pressure side to case or tank.

A charge pump, typically sized at 10–30 % of the main pump's displacement, makes up that loss. It draws from the reservoir through a suction strainer, delivers through a charge filter, and injects into whichever working line is currently at low pressure through a pair of check valves. A charge relief valve holds the low side at a controlled pressure, commonly of the order of 20–35 bar, which both suppresses cavitation at the pump inlet and supplies the control and servo circuits.

2.2 Why the loop is structurally hard to keep clean

The consequence of this architecture is that no filter sees the loop flow. A charge filter — the highest-performance element in most hostile-duty machines — filters only the make-up stream. Fluid inside the loop is exchanged only as fast as it leaves it — through the flushing valve and through internal leakage to the cases — and is replaced by filtered charge flow. If the flushing valve passes, say, 15 % of the loop flow, a particle liberated inside the pump rotating group that follows the flow has only about a 15 % chance of leaving each time it reaches the valve; on the flushing flow alone it passes through the motor on average six or seven times before it leaves, returning through the pump between one pass and the next. Leakage into the cases offers a second way out, at least for particles fine enough to pass through a clearance with it, and so shortens that count, though not to the single pass after which an open circuit would put the particle in front of a return filter. Every one of those passes is an opportunity to bridge a clearance.

This is the central asymmetry between closed and open circuits. In an open circuit, all returning fluid passes a return filter once per cycle. In a closed circuit, internally generated debris is trapped in the highest-value components in the machine and is removed only by dilution. It is the reason that internal wear in a closed loop is self-reinforcing, and the reason that suction-side or case-drain filtration alone is generally regarded as inadequate protection for a hostile-duty travel drive.

3. Clearances, particle size and the mechanism of wear

3.1 The critical interfaces

An axial piston unit is a stack of hydrodynamic and hydrostatic bearings operating in series. The three that determine its life are the slipper/swashplate interface, the piston/bore pair, and the cylinder barrel/valve plate face. Each is designed to run on a fluid film whose thickness is a small fraction of a human hair. Published design and analysis literature places the slipper film in the region of a few micrometers to roughly twenty micrometers depending on load and speed; the barrel/valve plate film in the low single micrometers up to around ten; and piston/bore radial clearance in the region of five to twenty-five micrometers. Control spool clearances are tighter still, often one to five micrometers radially, and rolling-element bearings in the same housing operate on elastohydrodynamic films of only a few tenths of a micrometer.

3.2 Mapping particles onto clearances

The damaging particle is not the largest one. Particles substantially larger than a clearance are excluded at the entry — they may dent, brinell or block, but they do not enter. Particles far smaller than the film pass through without contacting both surfaces. The population that does the work is the one comparable in size to the gap: it enters, bridges the film, and is dragged between two loaded surfaces. This is three-body abrasion, and because the interfaces of interest are in the 1–25 µm range, so-called silt — particles broadly in the 2–15 µm band, invisible to the eye and invisible in a settled sample — is the dominant wear agent in these machines.

Each abrasion event removes metal from both surfaces and typically generates two or more smaller particles. Those particles remain in the loop. This is why contamination-driven degradation in a closed circuit is autocatalytic rather than linear.

4. Coding fluid cleanliness

4.1 What an ISO 4406¹ code is

ISO 4406¹:2021 expresses a fluid's solid-particle content as three range numbers separated by slashes, corresponding to the cumulative counts of particles larger than 4 µm(c), 6 µm(c) and 14 µm(c) per millilitre. The suffix (c) denotes that the particle counter has been calibrated per ISO 11171² against a certified reference material; sizes are equivalent optical diameters, not physical dimensions, and are not interchangeable with the pre-calibration sizes used by older documents.

Each range number spans a factor of two in concentration, so a one-digit change is a doubling or halving. A code of 18/16/13 therefore means: more than 1 300 and up to 2 500 particles per millilitre above 4 µm(c); 320 to 640 above 6 µm(c); and 40 to 80 above 14 µm(c). The step-of-two structure is the most frequently misread feature of the system — an improvement from 21/19/16 to 18/16/13 is a factor of eight reduction in every size band, not a modest one.

4.2 Sampling is the weak link

A cleanliness code is only as good as the sample. Published sampling practice (ISO 4021⁵ for extraction from operating lines, ISO 3722⁶ for container qualification) exists because the sampling error routinely exceeds the analytical error. Samples must be drawn from a live, turbulent line rather than a drain or a reservoir surface; the sampling valve must be flushed; and the container must be verifiably cleaner than the fluid being measured. Light-extinction automatic counters per ISO 11500³ will miscount water droplets and air bubbles as particles, which is why an anomalously high count in a wet or aerated sample should prompt a microscopic count per ISO 4407⁴ before conclusions are drawn.

Typical published OEM guidance for axial piston units in closed circuits places the target in the region of 18/16/13, with tighter codes specified where electrohydraulic controls with fine spool clearances are present. Other classifications, such as SAE AS4059 (the current aerospace classification) and the obsolete NAS 1638 it superseded, use different size bands or class limits and are not directly convertible; conversion tables in circulation are approximations only.

5. Rating filter performance

5.1 Filtration ratio

Filter performance is expressed as a filtration ratio, β_x(c) = N_u/N_d, the ratio of the number of particles larger than size x upstream of the element to the number larger than x downstream, both counted per unit volume. Removal efficiency follows directly:

E = (1 − 1/β_x(c)) × 100 %

So β = 2 corresponds to 50 % removal at that size, β = 10 to 90 %, β = 200 to 99.5 % and β = 1000 to 99.9 %. A bare statement that an element is "10 micron" is meaningless without the β value and the calibration basis; "β_7(c) ≥ 200" is a specification, "10 micron nominal" is not.

5.2 How β is measured, and its limits

The reference method is the multi-pass test of ISO 16889⁷, in which test dust is injected at a controlled rate into a recirculating rig at constant flow while particle counts are taken upstream and downstream and the element is loaded to a defined terminal differential pressure. It yields β as a function of size and also the element's dirt-holding capacity. Companion standards cover fabrication integrity and bubble point (ISO 2942¹⁰), collapse and burst resistance (ISO 2941⁹), fluid compatibility (ISO 2943¹¹) and pressure-drop characteristics (ISO 3968¹²).

Two honest caveats apply. First, ISO 16889⁷ runs at constant flow; real drives see flow and pressure transients, and elements can release previously captured particles under such conditions — which is precisely why the cyclic-flow variant of the multi-pass method (ISO 23369⁸) exists, and why laboratory β should not be read as a guaranteed in-service efficiency. Second, β is measured with a specified test dust, not with the ductile metallic wear debris that a failing rotating group produces. Both effects tend to make field performance worse than the rating, not better.

5.3 Placement in a closed circuit

Charge-pressure filtration downstream of the charge pump is generally preferred over suction filtration: it permits a high-β element, tolerates a differential-pressure indicator, and cannot starve the charge pump as a clogging suction strainer can. Its limitation is that it protects only against ingested contamination, not against debris already resident in the loop. Case-drain filtration captures wear debris close to its source but must be sized for very low differential pressure, because case pressure limits on axial piston units are low — typically a few bar — and an over-restricted case line will lift the shaft seal. Offline (kidney-loop) filtration of the reservoir is the usual answer where a demanding target code must be held on a machine working in an abrasive environment.

6. Progressive efficiency loss and the case drain

6.1 The cube law

Laminar flow through a narrow rectangular gap of width w, length L and height h under a differential pressure Δp in a fluid of dynamic viscosity μ is, to first order:

Q = w h³ Δp / (12 μ L)

The cubic dependence on gap height is the single most important relationship in this subject. Abrasive wear that opens a clearance by only 26 % doubles the leakage through it. A clearance that has doubled leaks eight times as much. Volumetric efficiency, η_v = 1 − Q_leak/Q_theoretical, therefore falls slowly at first and then rapidly, and the machine's symptoms follow the same curve: a barely perceptible loss of tracking speed, then noticeably slower travel under load, then an inability to hold a slope or complete a swing.

6.2 Case drain flow as a condition indicator

Because essentially all internal leakage in an axial piston unit ends up in the case, case drain flow is a direct measurement of the aggregate of those clearances — provided leakage is all the case receives. Where the loop flushing valve discharges into the housing (§2.1), or in some designs the charge relief does, that flow must be excluded as the manufacturer's test conditions specify before the reading can be taken as leakage: at a flushing rate of the order illustrated in §2.2 it can exceed a new unit's leakage, and left in, it dilutes the change the cube law produces. New-unit case flow, measured that way, is commonly of the order of a few percent of rated flow, and many manufacturers publish a condemnation limit as a multiple of the new-unit value at stated conditions. Its diagnostic power comes from the cube law: with the non-leakage flow excluded, a clearance change too small to see on a bench is a large, unambiguous change in case flow.

The measurement is only meaningful when normalized. Case flow varies with speed, with the differential pressure across the unit, and inversely with viscosity — and viscosity of a typical mineral hydraulic fluid falls steeply with temperature, on the order of a quarter to a third per 10 K in the normal operating band. A case flow measured on a cold machine and compared against a limit specified at 50 °C will read low; the reverse comparison will read alarmingly high. Speed, load and temperature must all be recorded alongside the flow.

7. Water and aeration

Mineral hydraulic fluids dissolve water — typically a few hundred parts per million at ambient temperature, with the saturation point rising steeply as the fluid warms. Water above saturation appears as a free or emulsified phase, and the practical consequence is that a machine can hold water in solution when hot and precipitate it on cooling, so that free water is present at exactly the moment the drive is least protected. Water content is properly measured by coulometric Karl Fischer titration (ASTM D6304²¹); the visual "crackle" check detects only gross free water.

The recognized mechanisms of harm are corrosion of ferrous surfaces, hydrolysis of certain additive chemistries, promotion of varnish and sludge, and reduction of rolling-contact fatigue life in bearing steels — the last generally attributed in the literature to hydrogen liberated at the contact. The existence of the fatigue-life effect is well established, but it is not captured by the rating-life calculation of ISO 281¹⁸:2007: that standard's life modification factor takes lubrication condition through a viscosity ratio and contamination through a factor, e_C, defined for solid particles, so any allowance for water has to be judged separately. The magnitude is genuinely disputed: published reductions for water levels of a few hundred ppm range widely and depend strongly on the base fluid, the additive package, the steel and the test method. A responsible reading is that water at the hundreds-of-ppm level is unambiguously undesirable, and that any specific multiplier quoted for its effect should be treated as indicative rather than predictive.

Air behaves differently in its dissolved and entrained states. Mineral oil dissolves of the order of 8–10 % air by volume at atmospheric pressure, which is harmless. Entrained air — discrete bubbles — is not. It reduces the effective bulk modulus of the fluid dramatically at low pressures, making controls spongy and slow; it degrades the load capacity of hydrostatic bearings; and when a bubble is carried from the low-pressure side into the high-pressure side it is compressed near-adiabatically, producing local temperatures capable of thermally degrading the surrounding oil. This micro-dieseling is a recognized source of varnish and of the characteristic dark deposits found on valve plates. Fluid resistance to foaming and its air-release behavior are characterized by ASTM D892²² and ASTM D3427²³ respectively, and the system-side contributions — return lines discharging above fluid level, undersized suction lines, a leaking charge-pump inlet, an inadequate reservoir dwell time — are usually the dominant cause in the field.

Gaseous cavitation, in which dissolved air comes out of solution as local pressure falls below saturation, must be distinguished from vaporous cavitation, in which the fluid itself flashes below its vapor pressure. The damage differs: vaporous cavitation erodes surfaces and generates wear particles, while the air released by gaseous cavitation typically leaves surfaces darkened and varnished by the micro-dieseling described above rather than eroded. In a closed circuit both are usually symptoms of inadequate charge pressure, which the charge relief setting and the charge filter condition together determine.

8. Viscosity, film thickness and volumetric efficiency

Viscosity selection is a compromise between two failure modes. Too thin, and film thickness in the slipper and valve-plate interfaces falls toward the composite surface roughness, moving the contact from full-film into mixed and then boundary lubrication; leakage through every clearance also rises, since Q ∝ 1/μ. Too thick, and the charge pump cavitates on cold start, mechanical losses climb, and heat generation rises.

Axial piston manufacturers therefore publish an operating viscosity window, commonly an optimum band in the region of roughly 16–36 mm²/s with an absolute minimum around 10 mm²/s at the hottest point in the duty cycle and a cold-start maximum of the order of 1 000 mm²/s or higher. Because a machine may see both a winter start and a summer full-load duty, the fluid must satisfy both ends — which is the engineering case for a high viscosity index fluid classified as HV under ISO 6743-4¹⁶ and specified per ISO 11158¹⁷, with grade per ISO 3448¹⁵, viscosity measured per ASTM D445¹⁹ and index calculated per ASTM D2270²⁰.

One caveat deserves emphasis because it is frequently overlooked: viscosity index improvers are polymeric and shear. Multigrade hydraulic fluids exhibit both temporary viscosity loss under high shear rate and permanent loss as molecules are severed in service. A fluid that met its grade when filled may not meet it after a season, and the shear rates in the annular clearances of an axial piston unit are among the highest in the machine. Shear stability is a legitimate selection criterion, not a marketing distinction.

Film thickness and cleanliness are not independent variables. The specific film thickness λ = h/σ, the ratio of film thickness to composite roughness, sets how much of the load is carried by fluid rather than asperity contact; contamination sensitivity rises sharply as λ falls, because a thinner film means a larger fraction of the circulating particle population is now comparable to the gap. A drive running hot on a thin fluid is not merely less efficient — it is measurably more vulnerable to the same particle count it tolerated when cool.

9. Implications for repair practice

When a rotating group fails, it does not fail in isolation. The pump discharges its debris into the working line, through the motor, through the flushing valve, into the case volumes, along the case drain lines, into the cooler and into the reservoir. Hoses retain debris in their convolutions and in the crimp shoulders of their fittings; coolers retain it in low-velocity tube ends; the reservoir retains it wherever velocity is low. Every one of those reservoirs of debris is upstream of the replacement unit.

The engineering conclusion follows without needing field statistics to support it: replacing the failed unit alone reintroduces a new set of micrometer clearances into a circuit that is still full of the particles that destroyed the last set. Established practice after a debris-generating failure is to treat the whole circuit as contaminated — replace or thoroughly flush the working-line and case-drain hoses, replace all filter elements, drain and physically clean the reservoir rather than only draining it, flush or replace the cooler, flush the loop at a velocity sufficient to lift settled debris (turbulent flow is the criterion), replace the fluid, and verify by sampling to the target code before commissioning rather than after. A short run-in with a temporary high-efficiency element, followed by a confirming sample, is a proportionate final step on a high-value drive.

Two further points bear on repair. First, the loop flushing valve and the charge relief are frequently overlooked; a flushing valve that has stopped shuttling removes the loop's deliberate fluid-exchange mechanism, leaving internal leakage as the only exchange, and a charge relief set low or held open by debris removes its cavitation margin. Both should be verified, not assumed. Second, the diagnostic value of the case drain is greatest when a baseline exists. Recording case drain flow at defined speed, pressure and temperature at commissioning converts a later measurement from a judgment call into a comparison.

10. Limits of the present state of knowledge

Three areas remain genuinely unsettled and should be treated with corresponding caution. The quantitative relationship between an ISO 4406¹ code and component life is empirical and system-specific; life-extension multipliers quoted for a given improvement in code are useful as planning guidance but are not transferable across machine types or duty cycles. The magnitude of the water effect on rolling-contact fatigue, as noted in §7, varies widely across the published literature. And laboratory filtration ratios measured under constant flow with standardized test dust systematically differ from in-service performance under cyclic flow with ductile metallic debris; the direction of the error is known, the magnitude is not, and this is precisely why the cyclic-flow test method exists as a separate document. Practitioners are better served by trending — of cleanliness codes, of case drain flow, of operating temperature — than by any single absolute number.