Desktop version

Home arrow Health arrow Case Studies in Maintenance and Reliability: A Wealth of Best Practices


Testing the Hypothesis

According to our hypothesis, seal damage occurred principally at the time of starting the pumps. If true, we could accelerate seal failures by frequent changeovers. We discussed this procedure with Operations and they agreed to a daily changeover schedule for the trials.

Just prior to starting the standby pump, we planned to cool its stuffing box contents by using the cooled (seal) water from the running pump. For this purpose, we would connect the line from the cooler outlet of the running pump to its matching point on the standby pump. The flow through this line would be controlled carefully, as excessive flow would mean that the running pump would be starved and see an excessive rise in temperature. We designed and fabricated a new 1/2" pipeline to connect the outlets of the two coolers. It had isolation valves at each end, and a 1-mm orifice plate to restrict the flow (see Figure 40.2)


So far, we only had a theory. We would know if any of it was true when we started the trials. In the chart (Figure 40.3), we can see the result of the first trial carried out on March 8. At this time, the P1409A seal was brand-new,

Temporary Seal Piping Arrangement

Figure 40.2 Temporary Seal Piping Arrangement

Trial 1 Results

Figure 40.3 Trial 1 Results

while that in P1409B had been in-service for some time. Reading from left to right, the chart shows that P1409B was running and P1409A was about to start. The stuffing box water temperature of P1409B, which was normally at about 56°C, rose to 74°C. This was because the interconnecting 1/2" pipeline valves had been opened earlier, and the B pump seal's pumping ring was now supplying the A pump as well. In the stuffing box of the A pump, the water temperature, dropped from 108°C to 75°C.

This eight-minute printout shows the following:

Boiler Feed-Water Pump Seals 307

  • 1. The A pump was started about 100 seconds from the start of the print out. At this stage, the temperature of the water leaving the B pump's cooler was 75°C. Hence, the seals of both pumps were at 75°C.
  • 2. During the next 45-50 seconds, the A pump water temperature rose to 86.6°C. This 11.6°C rise was due to the heat generated at the seal faces.
  • 3. In the next three or four seconds, the A pump's seal pumping ring established a flow through its own cooler. The temperature then dropped to about 56°C. Thereafter, the temperature remained steady at this level.
  • 4. Once the A pump started and the B pump stopped, the interconnecting pipeline valves were closed.
  • 5. The B pump fluid temperature was about 74°C at the beginning of the chart. During the coast-down from full speed to zero, the heat generated by the seal faces due to viscous friction also fell.
  • 6. The pumping ring discharge temperature dropped by 11.6°C, but this took about 75 seconds, as the pump took longer to come to a stop.
  • 7. After the pump stopped completely, the temperature rose to reach the pump stuffing box temperature of 108°C, because at this time the interconnecting pipeline valves were also closed and there was no supply of cool water.

This chart shows that our hypothesis was not far off the mark. Note that the viscous seal friction produces a temperature rise of 11.6°C. The chart shows that the pump seal ran for about 45 seconds before its seal cooler became effective. Once this happened, the temperature dropped to 56°C, and remained there as long as the pump was running. Similarly when the pump was stopped, there was an initial fall in temperature by 11.6°C, confirming our evaluation of seal friction heat generation. Once the pump came to a halt,

Seal Performance on March 10-11

Figure 40.4 Seal Performance on March 10-11

the temperature rose gradually till it reached the pump suction conditions of 108°C.

The next chart shows the performance of the P1409B on March 10 and that of P1409A on March 11 after the changeover. As you can see, the B pump temperature profile was quite rough, while that of the A pump was smooth—note that these are 8-hour charts. The roughness in the B chart was because the seal faces were already damaged, causing erratic hot spots. As a result, steam is formed, leading to occasional seal separation. There was a clear rising trend in the temperature and the seal was close to failure. In any event, the B pump seal failed on March 12. The trial shows that the temperature profile was a good indicator of seal condition.

The daily change-over definitely accelerated the seal wear quite dramatically. This supports the hypothesis of seal damage due to dry running during pump starts.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >

Related topics