I recently conducted a failure analysis and a reliability audit on a 300-kilowatt hydrostatic transmission. This hydraulic system was running a synthetic-ester, biodegradable hydraulic fluid and the original set of pumps failed in less than 12 months.
The system was built and installed by a reputable firm. From a hydraulic engineering perspective, the circuit was adequately designed and the system well built. But from a maintenance and reliability perspective, there was plenty of room for improvement.
When I arrived on-site, I noticed that the hydraulic oil appeared dark in the sight glass, whereas the unused oil was the color of light honey. A check of the oil analysis reports indicated viscosity was increasing. I suspected oxidative failure of the oil and requested acid number and water content by Karl Fischer - tests which weren't included in the original test slate.
Failed Pumps
In the meantime, I turned my attention to analysis of the failed pumps. It was obvious the oil had been polymerizing for some time before the pumps failed. Internal components were heavily coated with a gum-like sludge (Figure 1). These deposits can block lubrication passages, reduce heat transfer and cause valve stiction.
The valve plate and cylinder barrel of both pumps exhibited damage from cavitation erosion - another possible side-effect of oxidative failure. The oxidation process typically diminishes foaming resistance and air release properties of the oil, which in turn causes damage through increased air entrainment and gaseous cavitation.
And The Results Are …
Acid number results came back at 9.5 mg KOH/g and Karl Fischer measured 2,200 ppm. For this particular lubricant, an acid number between 4 and 5 mg KOH/g triggers an oil change. Although water contamination was higher than desirable for this application, hydrolysis was ruled out as a significant factor in the formation of the sludge deposits, based on the lube manufacturer's experience.
All the evidence pointed to the fact that this hydraulic fluid, at $12 per liter, had suffered oxidative failure in less than 12 months. So with $20,000 worth of hydraulic fluid and $50,000 worth of pumps ruined in short order, the client was understandably concerned with what had gone wrong.
Suspecting high-temperature operation as the cause of the accelerated oxidation, I turned my attention to the historical operating parameters of the hydraulic system. The system was built with a temperature sender installed in the hydraulic reservoir, with alarm levels and shutdowns programmed into the system's programmable logic controller. Tests showed this instrumentation to be functioning properly. But the operators advised the system had never experienced a temperature alarm nor shutdown because of overtemperature.
This did not surprise me, because the temperature sender was in the wrong location. Let me explain. A hydrostatic transmission consists of a variable-displacement pump and a fixed or variable displacement motor, operating together in a closed circuit. In a closed circuit, fluid from the motor outlet flows directly to the pump inlet, without returning to the reservoir (Figure 2).
In addition to being variable, the output of the transmission pump can be reversed, so that both the direction and speed of motor rotation are controlled by the pump. This eliminates the need for directional and flow (speed) control valves in the circuit.
Because the pump and motor leak internally, allowing fluid to escape from the loop and drain back to the reservoir, a fixed-displacement pump called a charge pump is used to ensure that the loop remains full of fluid during normal operation.
In practice, the charge pump not only keeps the loop full of fluid, it pressurizes the loop to pressures between 110 and 360 PSI, depending on the transmission manufacturer. A simple charge pressure circuit comprised of the charge pump, a relief valve and two check valves, through which the charge pump can replenish the transmission loop (Figure 3). Once the loop is charged to the pressure setting of the relief valve, the flow from the charge pump passes over the relief valve and back to the reservoir.
Apart from losses through internal leakage, which are made up by the charge pump, the same fluid circulates continuously between transmission pump and motor. This means if the transmission is heavily loaded, the fluid circulating in this loop can overheat. To ensure the fluid in the transmission loop is exchanged with that in the reservoir and subsequently cooled, a flushing valve is installed in the circuit.
When the hydrostatic transmission is in neutral, the flushing valve has no function and charge pressure is maintained by the charge relief valve, usually located in the transmission pump. When the transmission is operated in either forward or reverse, the flushing valve operates so that charge pressure in the low-pressure side of the loop is maintained by the purge relief valve incorporated in the flushing valve. This purge relief valve is set approximately 30 PSI lower than the charge pump relief valve.
The effect of this is that cool, conditioned fluid drawn from the reservoir by the charge pump, charges the low-pressure side of the loop through a check valve located close to the transmission pump inlet. The volume of hot fluid leaving the motor outlet, which is not required to maintain charge pressure in the low-pressure side of the loop, vents across the flushing valve purge relief and back to the reservoir.
The important point here and what's relevant to this case is this: if the flushing valve malfunctions or is not configured correctly, there is no positive exchange of the fluid in the loop with that in the reservoir. This means the transmission loop can operate in overtemperature conditions while the fluid in the reservoir remains relatively cool.
As you can see, in a hydrostatic transmission, the correct location for the temperature sender on which overtemperature alarms and/or shutdowns are based, is in the transmission loop - not the reservoir.
A Failure of Maintenance or Design?
So was this case a failure of maintenance or a failure of design? It could be argued that it's both. A failure of design because if the temperature sender was correctly located in the transmission loop, the failure wouldn't have occurred. It was also a failure in maintenance because if the early warning signs of oxidative failure of the oil were picked up through observation and better test slate selection, the failure could still have been prevented.
