By Ray Beebe, MCM Consultants and Federation University
A previous article (Optimise overhauls of pumps to save energy, Pump Industry, October 2012) showed how quantified performance information can be used to optimise the time of an overhaul. Measurement of the balance device leakoff flow has long been recommended as a simple way of inferring the internal condition on multi-stage pumps with this feature. However this method may not always give the correct information on pump condition. Here Ray Beebe explains the balance flow method and discusses important considerations from his experience of it.
Igor Karassik, the late American pump guru, was often asked when a pump should be overhauled. His advice was that an overhaul is justified when the internal clearances were twice the design value, or when effective capacity has been reduced by about 4 per cent. However, before such a guideline can be used, a pump operator needs to correlate measured performance with as-found condition. A better way to balance the cost of overhaul against the cost of wasted energy was described in the earlier article.
Multi-stage pumps of split casing, ring-section design usually have all the impellers facing towards the suction end. To overcome the resulting axial thrust, some discharge flow is led through an annular clearance to act against a balance drum or balance disk. The resulting force is self-adjusting with pump flow or speed, and results in a smaller residual thrust loading on the bearings. Figure 1 shows such an arrangement.
Karassik also recommended that, where it applies the balance device leakoff, flow should be measured as a guide to pump condition and that overhaul should be considered when the flow has doubled.
This flow will increase as the annular clearance between the device and the pump casing increases with wear. Karassik said that it is therefore likely that the clearances of the wearing rings at each impeller will also have worn at the same rate. This is particularly so with the older generation pumps rotating at nominal 3,000r/min and having up to 11 stages.
The attraction of this method is that the balance leakoff line is quite small relative to the main flow line, and a permanent flow metering device is therefore not expensive. Devices such as Annubar™ flow elements have been fitted, relaying the flow to a panel meter or the control room.
At a power plant with several such variable speed pumps, permanent balance flow metering was read regularly and trended using a computer program originally bought for trending vibration data.
Initially, the plotted flows were erratic, until tests were run to find that the flow varied directly with the pump speed, the relationship being very close to linear (i.e. 20 per cent more speed, 20 per cent more flow). This means that for routine monitoring, the speed must be measured and the flow corrected to a standard speed value before trending.
Figure 2 shows the trend over some years of service. A nominal leakage flow of 17 L/s was selected as the indicator of the need for overhaul. As this flow is equivalent to 11 per cent of the duty flow, when balance device leakoff flow reached that level, an extra 250kW of power was being consumed. This is in addition to the power wasted due to internal recirculation, assuming that the wearing ring clearances were also worn.
Unfortunately, during the resulting overhaul the rings were replaced without measuring the clearances. As with all condition monitoring, it is important to find the actual condition so that this can be correlated with the predicted condition.
The method described in Optimise overhauls (October, 2012) could be used here to calculate when the investment in overhaul would be balanced by the savings in wasted energy.
Elsewhere, on another set of pumps of a multi-stage design, both head-flow and balance flow were measured for some years, but no correlation was found between the indications of condition.
Condition monitoring by performance analysis using head-flow was developed for another six pumps of a different type. These pumps had eleven stages, and operated at constant speed of nominal 3000 r/min. The head was measured using standard test quality pressure gauges, and the flow found using differential pressure measurements across the permanent orifice plate installed in the suction line. This flow element was provided to initiate the operation of the minimum flow device. Although it was not installed in the long straight pipe runs as required in flow standards, this achieved the repeatability required for condition monitoring.
During routine tests, one pump showed a test point well below the datum curve. The pump was duly dismantled, and with what I perceived as ill-concealed glee, the maintenance engineer reported that the interstage clearances were not worn. Our faith in the value of our tests suffered a credibility crisis!
A day later, the balance seat area was reached. It was found to be severely eroded, showing that water had flowed through the stationary annular gap between the balance device sleeve and casing, and left the pump behind the screws which retained the balance seat. Balance leakoff flow had obviously been very high, and if measured would have pointed to this damage. Extensive building up and machining of the casing was required. The pumps were modified to incorporate an O-ring seal on the stationary gap.
It seems that it would be unwise to rely on measurement of balance flow alone for condition monitoring. Both head-flow and balance flow should be measured for the best total picture of condition. Calculation of the optimum time to overhaul could be based on a higher than normal balance leakoff flow. The balance valve/sleeve/seat should be dismantled for inspection and replacement without opening the pump, where this is possible. Retesting afterwards would reveal whether the pump has to be fully dismantled for its interstage clearances to be restored.