INTRODUCTION
Ontario Power Generation (OPG) is the largest Canadian utility operating nuclear, thermal and hydraulic units. Currently, OPG is implementing a Predictive Maintenance (PdM) Program to monitor the health of plant components and equipment through application of three principle diagnostic tools; vibration monitoring and analysis, lubricant sampling and analysis, and infrared thermography. These Predictive tools will allow early detection of component and equipment problems so as to plan and schedule maintenance activities with minimal impact to plant operations or work schedules. One of the important elements of the PdM Program is to document each case and prepare a cost benefit analysis (CBA) based on the existing evaluation method developed by the Electric Power Research Institute. This analysis is performed to monitor the overall performance of the PdM program and to evaluate its effectiveness. This paper describes one of the earliest discoveries detected through the use of lubricant sampling and analysis at one of the OPG nuclear stations on the Auxiliary Boiler Feed Pumps (ABFP), which is a part of the Standby Safety System. It is an Operating Policy and Principle (OP&P) requirement requested by the Regulator that the pump is available when the heat transport system is above 90° C.

ASSESSMENT
The ABFP is a seven stage centrifugal pump with a balancing disk. It has split bushing, ring oil journal bearings. Tin babbitt is used as the bearing material. The bearing housing has an external cooling water system to control lube oil temperature. Oil is prevented from leaking out by elastomer seals. The manufacturer specified an optimum oil temperature in the range of 40° to 60° C. Inspection is required if the temperature reaches 85° C. The station operating bearing limit is set at 82° C.

During an outage, the ABFP is performance tested prior to return to service of a unit to ensure pump availability. As part of the performance test, the bearing oil is replaced with new oil. The ABFP is classified as a category one component in the Integrated Improvement Program (IIP). Maintenance activities associated with obtaining oil samples and performing vibration monitoring of those components are currently under development within the IIP Predictive Maintenance (PdM) program. These predictive measures will assist in the analysis for early detection of bearing failures.

Service water (lake water) is circulated even when the pump is shut down and, during winter, the inlet water temperature could be as low as 4° C, while in summer it could be as high as 25° C. The extensive cooling in the winter lowers the operating temperature of the bearings to below the optimum level masking the heat generated by the bearings. Also, during start up, highly viscous oil may not flow quickly enough to properly lubricate the rotating parts.

Vibration measurements were performed on this pump in May 1997. It was found that the vibration level on the pump bearing exceeded the recommended level. The vibration velocity spectra appear to indicate inadequate radial clearance between the impeller vanes’ outer diameter and the diffuser vanes’ inner diameter. It was recommended at that time that further vibration analysis be performed to determine the appropriate corrective measures.

On January 22, 1999, one of the station’s critical ABFPs, was checked and the oil replaced from both bearing housings as a part of a routine outage activity. Although the pump operated acceptably (as indicated by the pump’s on-line vibration probes and bearing temperatures were within the specification), the oil appearance was black and contained some solid particles.

During a Nuclear Unit outage in 1999, a standard performance test was performed before Unit start-up. The pump bearing temperatures were reported to be approximately 22° C. The vibration transducer installed on the pump was found to be defective.

Due to concerns about the blackish appearance of the oil, an oil analysis was performed. The oil analysis indicated that additives in the oil were depleted (oil due for replacement) and there was a high concentration of bearing wear material.

The oil was replaced with new oil and the pump test run. The bearing temperature measurement indicated that it stabilized approximately within 60 minutes. The Drive End (DE) bearing was at 24° C and the Opposite Drive End (ODE) bearing at 21° C. The motor bearing temperature was 35° C.

The cooling water inlet temperature was approximately 9° C and the outlet was approximately 22° C at the time of the performance test. The shaft temperature between the coupling and the inboard bearing was approximately 33° C. There was no evidence of high temperature spots or oil leakage.

The vibration spectrum, as measured by portable vibration measurement equipment, indicated low vibration levels (<3.7 mm/s), but also excessive bearing clearance and bearing concentricity problems.

The oil sample from the ODE bearing housing was black due to seal material contamination and the trace element analysis indicated the presence of bearing wear material. An oil analysis of the DE bearing provided no indication of bearing wear material. Therefore, the ODE bearing and the seals were replaced.

The bearing removed from the pump clearly showed signs of misalignment, as there were visible contact marks on the babbitt surfaces of the lower and upper pads at opposite ends of the bearings. There was an uneven thickness of babbitt layer due to wear. In addition, two axial cracks were observed on the lower ODE bearing pad. Those cracks, if allowed to propagate further, may have led to the break-up of the babbitt and possibly causing seizure of the bearing.

Both sealing faces of the oil seals were damaged. The material appeared very fragile and a residue of black powder was evident on the seals. See Photo 1 and Photo 2

RECOMMENDATIONS
An unexpected ABFP failure may have been averted by the early detection of the failed bearing. Maintenance strategies in place at the time for these pumps did not provide for early detection of bearing wear or failure.

The station chemical laboratory is presently not staffed to provide prompt analysis of the outage oil samples. The normal routine does not call for the collection of oil samples for the Aux. BFP due to the small amount of oil in the bearing housing. If a sample test is required, it is normally sent to an external laboratory for analysis.

A review of the maintenance records for other ABFPs concluded that other pumps might be at risk. Therefore, the following actions were recommended and implemented:

1. Implement as part of the PdM Program a lubrication sampling and analysis strategy for all ABFPs.

2. Obtain oil samples from each of the pump bearings and perform oil analysis for particle contamination and trace elements.

3. Review pump performance testing. Consider using portable vibration measuring equipment instead of on-line monitoring system.

4. Perform vibration analysis for all remaining ABFPs including high frequency envelope analysis for each pump motor set. The objective will be to determine which pump is suffering from radial clearance between the impeller vanes outer diameter and the diffuser vanes inner diameter. The analysis will also determine which bearing is defective and/or if the motor is suffering from an inherent electrical problem.

5. Review the pump bearing lubrication condition, particularly during cooler service water temperature conditions as follows:
a) Replace the mineral oil with synthetic fluid. A full analysis and manufacturer’s input would be required.
b) Review the oil change intervals.
c) Review the oil seal type and/or consider a more stable material.

Electric Power Predictive Maintenance Cost Benefit Analysis - Table 1

Electric Power Predictive Maintenance Cost Benefit Analysis - Table 1

COST BENEFIT ANALYSIS
A detailed cost benefit assessment of the ABFP was prepared to show the potential costs if the bearing failure had not been detected (see appendix for details). A risk-based approach was utilized in preparing the cost benefit analysis. This was accomplished by multiplying the probable cost of a possible outcome by the probability that the outcome would have occurred. The sum of the derived values produced the weighted value against the actual cost of the event (see table). This approach to the cost benefit assessment provides a conservative estimate of potential savings by predicting the failure before it actually occurs.

The following assumptions have been used in the appended assessment:

- Given that the failure mechanism has already been initiated on the seals and bearing babbitt, it is expected that the ABFP will become unavailable sometime in the future.

- The ABFP is part of the Station’s Standby Safety Support System. It is an operating license requirement that it is available when the reactor heat transport system temperature is above 90° C. The cost of non-compliance to the operating license has not been factored into the estimated savings.

- For failure mode “c”, no cost of reactor shutdown is included. It is assumed that the failure is gradual and the repairs can be preplanned. Otherwise the reactor will need to be shut down because repairs can not be completed within eight hours as required by the station safety procedure.

- For failure mode “a” and “b”, it is assumed that once the Aux. BFP is found to be unavailable, technical assessment of the failure is done quickly and the corrective action is completed within about three shifts. It is also assumed that spare parts are readily available at the station.

- It is assumed that failure occurs during off peak time. The cost of replacement power will be much higher if the failure occurs during peak power demand.

CONCLUSIONS
Had the situation gone undetected and if the pump had been placed in service, failure of the pump motor assembly may have occurred. Based on the risk-based cost benefit analysis, the economic impact of such an incident is estimated to be over $130,000.

This case exemplifies how a properly implemented condition monitoring improves the quality of machine maintenance decisions and helps the organization to meet its safety and economic goals.
Acknowledgment is extended to the Maintenance Department, Chemical Laboratory and Engineering Staff in providing prompt and efficient responses to this case and adverting delays to the planned outage schedule.