Ester Fluid Failure: How a Hydrostatic Pump Loss Was Traced

Based on Casey's Fluid Power Journal case (2013), a 300 kW ester hydrostatic drive failed within 12 months from undetected loop overheating that oxidized the fluid. The finding holds for this closed-circuit transmission, whose alarm and shutdown temperature sender sat in the reservoir. No loop temperature data were published, so overheating is the investigator's inference.

• The original pumps on a 300 kW closed-circuit drive running a synthetic ester, biodegradable fluid failed in less than 12 months (Casey, Fluid Power Journal, 2013).

• The 2013 article reports an acid number of 9.5 mg KOH/g, against an oil change trigger of 4 to 5 mg KOH/g for that fluid (Casey, Fluid Power Journal, 2013).

• The 2016 retelling of the same case gives an acid number of 10 and 6,500 ppm water, where the 2013 article gives 2,200 ppm (Casey, Power and Motion, 2016).

• One hydrostatic motor maker applies its temperature limits at the hottest point, normally the motor case drain, with a recommended range of 60 to 85 °C (Danfoss, 2021).

This analysis applies to closed-circuit hydrostatic transmissions with a charge pump and a loop flushing valve running synthetic ester fluid (category HEES). It does not cover open-circuit systems, where return oil passes through the reservoir, and it does not settle where the water in this case came from, which neither published version identifies.

What happened to the ester hydrostatic pump?

The ester hydrostatic pump failure happened on a 300 kW hydrostatic transmission running a synthetic ester, biodegradable hydraulic fluid, where the original set of pumps failed in less than 12 months (Casey, Fluid Power Journal, 2013). Brendan Casey ran the failure analysis and reliability audit. He judged the circuit adequately designed and well built from an engineering view, with plenty of room for improvement in maintenance and reliability.

On site, the oil looked dark in the sight glass, while unused oil was the color of light honey. The existing oil analysis reports showed viscosity increasing. Casey suspected oxidative failure and requested two tests that were not on the original test slate: acid number and water content by Karl Fischer titration.

Internal parts of the failed pumps were heavily coated with a gum-like sludge, which Casey read as evidence that the oil had been polymerizing for some time before the pumps failed. The valve plate and cylinder barrel of both pumps also showed cavitation erosion, which he linked to oxidation through reduced foaming resistance and air release (Casey, Fluid Power Journal, 2013).

The operators reported that the system had never raised a temperature alarm or an overtemperature shutdown. Its temperature sender, alarm levels and shutdowns tested as working. Both the fluid charge and the pumps were written off. The same investigator retold the case in a 2016 blog post with different test values, compared in a later section of this article (Casey, Power and Motion, 2016).

What could explain the ester hydrostatic pump failure?

Five candidate causes could explain the ester hydrostatic pump failure, and the published evidence supports loop overheating, enabled by poor loop flushing, as the main driver. The other three, water hydrolysis, mineral oil carryover and a wrong viscosity grade, have general evidence for them, but none fits the oil and the damage as well. The table below sets each candidate against the evidence for and against it from the two published versions of the case and from field research on biodegradable hydraulic oil:

The decisive row is loop overheating: the acid number of 9.5 mg KOH/g was roughly double the top of the 4 to 5 mg KOH/g change trigger for that fluid (Casey, Fluid Power Journal, 2013), and the sludge matches polymerized oxidation products. The flushing valve row is not a rival cause but the likely route to the overheating, and water could have added to oxidation without being the main driver (Danfoss, 2016).

Acid number alone cannot separate the two chemical candidates, because hydrolysis of synthetic ester bio-oil also raises the acid value (Ohkawa et al., AIST, 2019). What separates them in this case is the pattern: dark color, rising viscosity and gum-like deposits point to oxidative polymerization rather than to hydrolysis. A hydrolytic stability result for the fluid, or a record of earlier fluid changes and top-ups, would have tested the water and carryover rows directly, and neither version reports one.

What did the ester fluid acid number and water results show?

The ester fluid acid number and water results showed severe oxidation and more water than desirable: 9.5 mg KOH/g and 2,200 ppm in the 2013 report, against a 4 to 5 mg KOH/g oil change trigger for that lubricant (Casey, Fluid Power Journal, 2013).

Acid number counts acidic constituents, including oxidation products formed in service. The current ASTM method states: "It is also sometimes used as a measure of lubricant degradation in service. Any condemning limits must be empirically established." (ASTM International, D664-24, 2024). That is why the 4 to 5 mg KOH/g trigger belongs to this fluid only and should not be carried over to another ester or to mineral oil. Neither version of the case names the acid number test method.

Karl Fischer titration measures total water and cannot separate dissolved from free water (Danfoss, 2016). Coulometric Karl Fischer covers water from mg/kg up to percent levels in lubricating oils (ASTM International, D6304-25, 2025). The 2,200 ppm reading is about three times a 700 ppm critical value that one pump maker gives as a rough guide for HEES fluids (Danfoss, 2016). The AIST study reports no hydrolysis where water entering excavators stays at 0.1 % or less, and cites hydrolysis at a few percent (Ohkawa et al., AIST, 2019), so 2,200 ppm sits between those points. Casey reports hydrolysis was ruled out as significant on the lubricant maker's experience, not a test. Both results are single points taken after the failure, so no trend exists.

Why did the ester fluid oxidize so fast in a closed loop?

The ester fluid oxidized fast in the closed loop because the same fluid circulates between pump and motor, and without working flushing flow the loop can overheat while the reservoir stays cool (Casey, Fluid Power Journal, 2013). Motor outlet fluid flows straight to the pump inlet, and a charge pump replaces fluid lost to internal leakage and pressurizes the loop to about 7.6 to 24.8 bar (110 to 360 psi), depending on the transmission maker (Casey, Fluid Power Journal, 2013).

The flushing valve is the exchange point: its purge relief, set approximately 30 psi (about 2 bar) lower than the charge relief valve, vents hot motor outlet fluid to the reservoir (Casey, Fluid Power Journal, 2013). If it malfunctions or is set wrongly, there is no positive exchange.

Commercial bio-oils made for strict environmental rules have lower oxidation stability than petroleum hydraulic oil, because hindered phenol and aromatic amine antioxidants and zinc dialkyldithiophosphate (ZDTP) cannot be used for aquatic toxicity reasons (Ohkawa et al., AIST, 2019). As a rule of thumb, a rise in fluid temperature from 80 °C to 90 °C may cut fluid life by 50 % (Danfoss, 2016), and an axial piston pump cylinder block can run 30 °C or more above the oil, reaching 110 °C at 80 °C oil (Ohkawa et al., AIST, 2019).

Oxidation can then feed itself: oxidized oil loses foaming resistance and air release (Casey, Fluid Power Journal, 2013), and compressed bubbles create local hot spots that heat the surrounding bio-oil (Ohkawa et al., AIST, 2019). This suggests a self-reinforcing cycle.

Why did the reservoir temperature sensor miss the loop oxidation?

The reservoir temperature sensor missed the loop oxidation because, in a closed-circuit hydrostatic transmission, loop fluid flows from the motor outlet directly to the pump inlet without returning to the reservoir (Casey, Fluid Power Journal, 2013). The reservoir sender drove alarm levels and shutdowns in the programmable logic controller (PLC). It tested as working, yet the operators reported no temperature alarm or shutdown in the life of the system.

Casey's conclusion is direct: in a hydrostatic transmission, the sender on which overtemperature alarms and shutdowns are based belongs in the transmission loop, not the reservoir (Casey, Fluid Power Journal, 2013). Component maker documentation points the same way. For one series of hydrostatic motors, "The high temperature limits apply at the hottest point in the transmission, which is normally the motor case drain." (Danfoss, 2021). The same document notes that case drain fluid is typically the hottest fluid in the system, and gives a recommended range of 60 to 85 °C and a rated limit of 104 °C at that hottest point (Danfoss, 2021).

A reservoir reading inside its alarm band therefore says little about loop temperature when flushing flow is weak. The two locations do not conflict: Casey names the loop, the motor maker names the case drain, and neither is the reservoir. Casey judged the case both a design failure, because a correctly located sender would have prevented it, and a maintenance failure, because oil color and a better test slate could have caught the oxidation earlier (Casey, Fluid Power Journal, 2013, "A Failure of Maintenance or Design?").

Why do the two published versions of the pump failure differ?

The two published versions of the pump failure differ in the acid number, the water content, the sequence detail (2016 adds that the pumps failed shortly after the fluid) and the depth of the root cause, and neither text explains why (Casey, Fluid Power Journal, 2013; Power and Motion, 2016). The 2016 post on Power and Motion is shorter and retells the case for a maintenance audience without the circuit analysis. The table below compares the two versions item by item:

The largest gap is water: 6,500 ppm in 2016 against 2,200 ppm in 2013 (Casey, Power and Motion, 2016), a difference of about three times that rounding cannot explain. The acid number gap of 9.5 against 10 could be rounding. Unconfirmed explanations include a second sample taken later, after more water entry, or numbers recalled from memory three years on. Neither version gives a sample date.

Both versions agree on the core facts: a 300 kW hydrostatic transmission, a synthetic ester biodegradable fluid, failure inside 12 months and an oil analysis program that missed the oxidation (Casey, Power and Motion, 2016). The conflict is in the numbers and the depth of explanation, not in the story.

For decisions, the 2013 article is the stronger record: it is the earlier and fuller account, names the water test, gives the change trigger and explains the circuit. Anyone citing the 2016 values should state that the earlier account reports lower figures.

Conclusion and certainty on the oxidation cause

The oxidation cause is well supported, while loop overheating as the driver is a reasoned inference, and the role of water is unresolved (Casey, Fluid Power Journal, 2013). Sorting each claim by how it is known shows where the case is solid and where it rests on the investigator's judgment.

Confirmed by reported observation or test:

• The fluid suffered oxidative failure: dark oil, rising viscosity, gum-like sludge on pump parts and an acid number of 9.5 mg KOH/g against a 4 to 5 mg KOH/g trigger (Casey, Fluid Power Journal, 2013).

• The temperature sender for alarms and shutdowns sat in the reservoir, worked when tested and never alarmed (Casey, Fluid Power Journal, 2013).

• Both pumps showed cavitation erosion on the valve plate and cylinder barrel (Casey, Fluid Power Journal, 2013).

Inferred from the circuit and the oil:

• Loop overheating drove the oxidation. Casey states he suspected high-temperature operation, and the circuit logic supports it, but no loop temperature was recorded.

• A malfunctioning or wrongly configured flushing valve allowed the overheating. The 2013 article presents this as the relevant mechanism, not as a finding from inspecting the valve.

Unconfirmed:

• Where the water came from, and whether 2,200 or 6,500 ppm is the representative value (Casey, Fluid Power Journal, 2013; Power and Motion, 2016).

• Whether water contributed to oxidation; hydrolysis was excluded on the lubricant maker's experience, not by a test.

• Operating hours, so the fluid life cannot be compared with the 3,000 h change interval that one study sets for standard European saturated synthetic ester bio-oils (Ohkawa et al., AIST, 2019).

Action and verification for ester fluid loops

Action on an ester fluid loop with this failure pattern starts at the temperature measurement and ends with a verified oil trend, because the case failed through a misplaced sensor and a test slate without oxidation tests (Casey, Fluid Power Journal, 2013, "A Failure of Maintenance or Design?").

1. Add a temperature sender in the loop or the motor case drain and alarm against the component maker's limit for the hottest point, for example 60 to 85 °C recommended for one motor series (Danfoss, 2021), or the fluid maker's lower limit for HEES ageing, whichever is lower. Verify by logging loop and reservoir temperatures under heavy load; a persistent gap suggests low flushing flow, to confirm on the valve.

2. Check the flushing valve: flow present in forward and reverse, and purge relief set below charge relief, approximately 30 psi lower in the case circuit (Casey, Fluid Power Journal, 2013). Do not raise flushing flow beyond the maker's data; excessive motor loop flushing flow may prevent the system from building required pressure in some conditions (Danfoss, 2021).

3. Add acid number and Karl Fischer water to the test slate and trend both against the fluid maker's limits (Casey, Fluid Power Journal, 2013).

4. Replace the oxidized fluid. Drain and handle it following the fluid's safety data sheet (SDS) and the used oil rule that applies where you work, such as 40 CFR Part 279 in the US, which defines used oil to include used synthetic oil (US EPA, eCFR, 2026). Drain housings completely to avoid mixing fluid types (Danfoss, 2016).

5. Find and stop the water entry before refilling.

Lessons as an ester fluid loop checklist

The main lesson of the ester fluid loop case is that a closed hydrostatic loop needs its own temperature and oil condition checks, because the reservoir and a particle count can both look normal while the fluid oxidizes (Casey, Fluid Power Journal, 2013; Power and Motion, 2016). The checklist below turns the case into routine checks for any closed-circuit drive running synthetic ester fluid:

The loop temperature row matters most: Casey judged that a sender in the transmission loop would have prevented this failure, which the reservoir sender could not (Casey, Fluid Power Journal, 2013, "A Failure of Maintenance or Design?"). Three further lessons follow. A test slate built around particle counts misses oxidation, so acid number and water belong on it from the first sample. Ester fluids give less margin on temperature than petroleum oil, so the loop limit should be set with the fluid maker, not copied from a mineral oil system. And when a published case is used as evidence, the earliest full account should be checked against later retellings, because the numbers can change. Lubinpla works cases like this from a customer's own TDS, SDS and site history and returns one cited document for review.

Lubinpla applies this kind of reasoning to your own documents and site history and drafts the document for your review before it goes out.

Sources checked 2026-10-13.

Sources

1. Fluid Power Journal. Brendan Casey. Analysis of a Hydraulic Oil Failure. 2013-02-28. https://fluidpowerjournal.com/hydraulic-oil-failure/

2. Power and Motion. Brendan Casey. The Anatomy of a Hydraulic Maintenance Failure. 2016-11-01. https://www.powermotiontech.com/hydraulics-at-work/article/21886639/the-anatomy-of-a-hydraulic-maintenance-failure

3. AIST. Ohkawa S., Yun S., Hibiya T., Nishimura H. A new process to develop a hydraulic system adapted to biodegradable hydraulic oil for construction machinery. Synthesiology Vol. 12 No. 2 (2019), English edition 2021. https://www.aist.go.jp/pdf/aist_e/synthesiology_e/vol12_no2/vol12_02_p56_p74.pdf

4. Danfoss. Technical Information: Series 90 Axial Piston Motors, BC152886483265. November 2021. https://assets.danfoss.com/documents/192360/BC152886483265en-000711.pdf

5. Danfoss. Technical Information: Hydraulic Fluids and Lubricants, 520L0463. July 2016. https://assets.danfoss.com/documents/latest/54030/BC152886484524en-000902.pdf

6. ASTM International. D664-24 Standard Test Method for Acid Number of Petroleum Products by Potentiometric Titration. 2024. https://www.astm.org/d0664-24.html

7. ASTM International. D6304-25 Standard Test Method for Determination of Water in Petroleum Products, Lubricating Oils, and Additives by Coulometric Karl Fischer Titration. 2025. https://www.astm.org/d6304-25.html

8. US EPA. 40 CFR Part 279 Standards for the Management of Used Oil. eCFR, current as of 2026-10-08. https://www.ecfr.gov/current/title-40/chapter-I/subchapter-I/part-279

Written by the Lubinpla team from the sources listed and checked against them before publication.