Vibration Points to Piston Pump Failure

Research project uses vibration analysis to accurately predict failure of new and rebuilt variable-displacement hydraulic-piston pumps.

By Garrett Wolfe, CAT III Vibration Analyst, Vallourec Star

Unplanned downtime, particularly in facilities that operate around the clock, must be avoided, even if automated backup systems are in place. At our Vallourec Star automated tube mill plant in Youngstown, OH, we are determined to find ways to improve reliability and predictability. Our most recent effort involved predicting variable-displacement hydraulic- piston-pump failures.

Catastrophic failure in these pumps commonly occurs at the piston-shoe assembly that connects the pistons to the swashplate. The most common “predictor” has been leakage, which is a less-than–desirable indicator. As part of our reliability effort, I sought to use vibration analysis to provide a more consistent, earlier indicator of pending failure.

My work involved 18 months of data collection, analysis, and pump-tear-down inspections. That was followed by more than 24 months of using the resulting vibration-analysis method on more than 20 pumps to predict failure with a 100% success rate.

Pumps and Procedures

Our plant uses nine-piston variable-displacement axial-flow piston pumps that operate at 1,200 rpm and a pressure of 2,174 psi (150 bar). The pumps use water glycol (FR46) in a controlled temperature range of 100 to 112 F. As many are aware, life expectancy of pumps using water glycol is approximately 25% of a unit using mineral oil.

Piston failure typically occurs where the piston and the shoe are crimped together. The piston is factory assembled, free of end play, and has a maximum tolerance of  0.0027 in. (0.07 mm). Generally, the pumps will continue to run without piston failure until the end play is 0.006 in. (0.152mm).

The swashplate in these pumps is in a constant state of change during operation. To put the swashplate in its most stable state prior to taking a reading, the pump is placed in a deadhead condition.

If a pump has the high-pressure cradle-bearing-flush option, a high-pressure valve will need to be installed to prevent the flush line from skewing the vibration data. When testing, that valve must be closed and the amount of time the pump is in the deadhead condition minimized to prevent heat buildup. Data only needs to be collected at the inboard axial position on the high-pressure side of the pump.

New Pump Reading

Rebuilt Pump Reading

These spectra show readings from a new pump (top) and a rebuilt pump (bottom). Note the different amplitudes of the 18th order. The new pump has an amplitude of 0.0240 in./sec., and the rebuilt pump has an amplitude of 0.0572 in./sec.

Pump Ready For Rebuild

This spectrum indicates that the pump needs rebuilding. The 18th order has an amplitude of 0.100 in./sec. This pump had previously been rebuilt, therefore the rebuilt-pump analysis was used. The piston-to-shoe end play of all nine pistons was between 0.004 in. (0.102 mm) and 0.005 in. (0.127 mm). Manufacturer limit is 0.0027 in. (0.07mm).

Data Collection

When setting up the analysis software for data collection, set the FMAX to 70.5 orders, 1,600 lines of resolution, and reading the inboard axial position on the high-pressure side of the pump. FMAX is set at 70.5 orders to view the piston harmonics and other normal fault indicators.

To analyze the collected data, set the spectral view to velocity and orders. The force acting on the piston as it changes direction is the factor that needs to be analyzed. As there are nine pistons in a pump and the impact occurs twice per revolution per piston, the focus is on the 18th order.

Note when establishing baseline readings whether those readings are on a new or rebuilt pump as this will have some affect on the amplitudes (view spectral examples). The difference in amplitudes will depend on what was replaced at the time of rebuild. As an example, if the pistons were replaced but the drum assembly was reused, there will be a difference in amplitude of the 9th order and its harmonics, including the 18th order, due to the amount of wear on the drum, even though it is still within bore specifications.

Catastrophic failure in variable-displacement hydraulic-piston pumps commonly occurs at the piston-shoe assembly that connects the pistons to the swashplate.

Data Analysis

New-pump analysis: Since there will already be 18th-order indicators showing on the spectrum, which is a harmonic of the 9th order, of a new pump, subtract the 9th-order amplitude of the baseline reading from the 9th-order amplitude of the current reading and divide by two. Subtract the 18th-order amplitude of the baseline reading from the 18th-order amplitude of the current reading. Now subtract the 9th-order conclusion from the 18th-order conclusion. If the calculated rate of change exceeds 0.025, the pump will need to be rebuilt.

Note that the calculations for a new pump will only be accurate when all of the criteria listed previously are the same. Changes to those criteria will alter calculation results. Therefore, the calculations are to be used only as a beginning reference point for system evaluation.

Rebuilt-pump analysis: Rebuilt pumps can and most likely will have increased wear on reused components, causing a significant change at the 9th order, therefore causing change to the 18th order. On a rebuilt pump, look for the 18th order to reach
0.1 in./sec. At this point, the pump will need to be rebuilt. If 18th-order amplitudes reach 0.12 in./sec. the pump is running the risk of piston failure. Look for the 9th order to have sidebands at the turning speed, indicating excessive wear on other linked components, such as the drum.

By allowing a pump to run to failure, there will be a rather costly rebuild averaging $11,000 to $12,000, provided the unit can be repaired. Using vibration analysis, we can predict failure as far as two weeks out, depending on the number of starts and shut downs the pump has experienced. By pulling a pump at the two-week-out point, we have realized an average repair cost reduction of 60% to 65%. This approach has also prevented contamination of the entire hydraulic system (tanks, other pumps, lines, hoses, and control-valve assemblies), resulting in a significant increase in the expected life span of those components. EP

Garrett Wolfe is a CAT III vibration analyst at Vallourec Star, Youngstown, OH. He holds ISO/ ASNT certifications in ultrasound and infrared technologies and is responsible for condition-based motor lubrication and all rotating and non-rotating equipment alignment activities at the plant. Prior to joining Vallourec, he spent 20 years performing vibration-analysis functions in the automotive industry. Vallourec Star is a leading producer of premium seamless pipes used primarily in oil and gas applications. The company offers the latest technology in steel-making and pipe-mill production, heat treatment, threading facilities, and customized specialty-service products.