Optimization of pump maintenance
Condition based repair of piston pumps using online diagnosis

Permanent condition monitoring of pumps contributes significantly to increased reliability and efficiency as well as reduced maintenance costs of plants. This article describes an acoustic online diagnosis system for piston pumps, which allows the detection of valve damages in an early state. Maintenance can be optimized with this diagnosis system.

In the past years, demands for cost saving in chemical engineering facilities increased considerably. This also contributed to higher demands on safety standards, efficiency, reliability, availability, and diagnosis means of operated pumps. These criteria are directly related to costs of downtime production loss, replacements, maintenance, and repair.
Online condition monitoring is a method allowing condition related repair providing increased time of exploitation, reduced downtime and related costs.
Considering piston pumps (Fig. 1), only the development of an acoustic online monitoring system for pump valves allows condition related maintenance. It is well known that availability and efficiency of piston pumps are extensively dependent on the condition of their delivery valves. Leaks resulting from solids in the sealing area or from wear at valve seats may affect the output of pumps or even cause the failure of the entire unit. The acoustic online monitoring system described in this article allows the detection of leakages in valves in an early state.

Diagnosis by analysis of acoustic emission
Leaks are usually in the turbulent regime and particularly in the high pressure range cavitation is present. This produces sound and ultrasound, which can be detected on the outside of the valve with a
suitable sensor. The sound level is a measurement for leakage. Demands on the diagnosis system are:
-observation of all delivery valves of the pump,
-detection of leakages in an early state,
-independence from the sound generated by normal operation of the pump,
-independence from sound generated in the plant,
-operation in a wide range of pump stroke frequency,
-operation in a wide pressure range,
-tolerance for temporary leakages due to short-term capturing of particles in the
sealing area,
-simple installation,
-simple parameterization,
-generation of diagnostic information from the monitored sound emission,
-in addition to valve monitoring, monitoring of diaphragms and oil temperature.
The online diagnosis system Sitrans DA (basic concept and test in co-operation of Siemens and Feluwa, development by
Siemens) meets these requirements. Its acoustic emission sensors are installed on the outside of the valves and are optimized for high sensitivity for leakage signals
while being insensitive for pump operation signals. A suitable signal processing realizes the automatic adaptation of measurement to changing load conditions, such as pressure or number of strokes. Only the alarm limits have to be set by the operator.
Optimization of maintenance
Until now, leakage or failure of delivery valves only became evident by insufficient pump performance, e.g. reduced flow rate. Actual reasons for such loss then had to be found out arduously, sometimes by removal and revision of all delivery valves. The loss of efficiency due to such a damage was accepted.
Service life characteristics of delivery valves of a hose-diaphragm piston pump are illustrated in Fig. 2. In phase 3, valve condition declines due to wear and a repair is required. Valve wear results in
leaks and proceeds progressively. In its initial stage, only the valve seat is damaged, later on the ball as well. The time between initial wear and valve failure depends on the actual duty of the valve, the working pressure and the conveyed fluid. It may range between 5 and 60 days. The transition from phase 2 to phase 3 is determined by the valve load and can be accelerated e.g. by short-term capturing of particles in the sealing area. For this reason it is hardly possible to precisely predict the time of valve failure and if all valves of a pump will be subject to the same wear.
Maintenance of a pump can be based on one of the three strategies:
-time-based,
-corrective,
-condition based.
The suitability of a strategy depends on the plant and finally the availability of condition monitoring systems.
Time-based maintenance does not require condition monitoring. Wear parts are replaced before failure. Too short repair intervals result in high costs due to frequent repair. However, unexpected shutdowns are avoided and plant downtime is scheduled.
Corrective maintenance allows the operation of the unit up to its breakdown. The main disadvantage is the unplanned shutdown for repair.
Condition based maintenance requires a possibility for monitoring the condition of the plant or unit. Its advantages are:
-damages are detected before breakdown of the unit,
-shutdowns for repair can be planned,
-repair can be done before expensive subsequent damages arise,
-service life of components can be fully exploited,
-reduced downtime, since searching the location of the damage is not required,
-automatic inspection of the plant or unit.
Field tests of the online diagnosis system Sitrans DA, which started one year ago, proved the reliable detection of damages of delivery valves. Fig. 3 shows an example for the detection of a leakage in a Feluwa triplex hose-diaphragm piston pump. The conveyed fluid is a suspension with a pressure of 150 bar. The measurement shows a sudden increase of the measured sound level at the suction valve of head 2 well above the alarm limit. About two hours later the pump was shut down and the valve replaced. Then the measured value returns to small values. Considering the fact, that this pump can be operated up to two weeks after the first detection of damages, the online diagnosis system provides sufficient time for planning the shutdown.
Fig. 4 shows a life-cycle consideration of investment and maintenance costs of three different pumps of equal performance. Accumulated costs over 30 years can be reduced to 50% with the help of the online monitoring system. Furthermore, the efficiency losses explained in Fig. 5 have to be considered.
Summary: The test of the online diagnosis system Sitrans DA demonstrated the reliable detection of leakages of delivery valves of piston pumps in an early state. This offers significant advantages in maintenance since inspection is done automatically and shutdown for repairs can be planned in advance. In some cases even the installation of standby pumps becomes unnecessary. Thus the availability of an online diagnosis system allows the predictive and economical maintenance of pumps.
This monitoring principle is not only limited to the delivery valves of pumps, but can also be applied to leakages in delivery valves of compressors, control valves, on/off valves etc.