Docstoc

Performance monitoring and analy

Document Sample
Performance monitoring and analy Powered By Docstoc
					           Experiences in Pump Condition
           Monitoring by Performance Analysis
                            Ray Beebe                                                                       The availability of non-intrusive flow
                            School of Applied Sciences and                                            meters has greatly facilitated field tests
                            Engineering, Monash University                                            where access to a length of pipe is available
                                                                                                      If use on lined pipes is proposed a trial is
                                                                                                      suggested before buying one. Some plant
                           This paper was originally presented at the International                   installations may not have any available
                           Conference of Maintenance Societies, ICOMS2007.                            pipe, and other methods are needed.
                                                                                                            A suitable tank of uniform dimensions
                                                                                                      may be available in the system, and fittings
           Predicting maintenance requirements of pumps by condition                                  and pipe valving arranged so that its change
Abstract




                                                                                                      in level with time can be measured and the
           monitoring to detect internal wear is a well-established
                                                                                                      flow rate calculated. Allowance is required
           procedure. The head-flow test is the basic method, and is also                             if a tank to be used is on the suction side of
                                                                                                      the pump, as the suction head will decrease
           relevant to investigation of suspected poor performance. The
                                                                                                      as the tank is emptied. If the discharge head
           head is usually readily measured, but the plant layout may                                 is very high in proportion it may be that a
           mean that measuring the flow is difficult or expensive. Where                              change in suction head is negligible, as long
                                                                                                      as adequate NPSH (Net Positive Suction
           a suitable vessel exists in the system, a simple way is to measure                         Head) is maintained. This method is
           the change in volume with time. This paper describes some                                  applicable for open tanks with that benign
                                                                                                      liquid, water, at atmospheric conditions.
           experiences with the testing of pump performance, including                                Sealed tanks containing less benign
           measuring the flow of some slurry pumps that pumped from                                   liquids such as hydrocarbons may have
                                                                                                      a manometric level indicator than can be
           an underground sump of non-uniform dimensions. The volume                                  tried. This paper describes an adaptation of
           change method was successful, using another tank in the system.                            this method where the suction chamber in
                                                                                                      the system was not of uniform dimensions.

       KEYWORDS: pumps, condition monitoring, flow measurement
                                                                                                      THE SYSTEM
                                                                                                           A somewhat simplified description of
                                                    INTRODUCTION                                      the relevant parts of the system, a closed
                                                                                                      cycle in a coal fired power station, is as


                                                   P
                                                           erformance monitoring and analysis         follows –
                                                           is often applied to pumps to detect           Sluice pumps injected jets of water
                                                           and monitor the extent of internal               along the ash sluices from the boiler
                                                    wear. Of the various methods available, that            hearth and the dust collection plant
                                                    of head-flow is preferred [1] as it reveals             to flush the slurry along into the ash
                                                    the condition of the pump and also of the               sump.
                                                    system it serves. Other methods may also
                                                                                                          The ash sump was below ground
                                                    be appropriate, such as vibration analysis to
                                                                                                           level, was of non-uniform cross-
                                                    monitor bearings, looseness, unbalance and
                                                                                                           section, and the contents were barely
                                                    alignment.
                                                                                                           visible.
                                                         Head has long been readily measured
                                                    with standard test pressure gauges, with a            The ash disposal pumps conveyed
                                                    wide range of electronic transducers available.        water with low ash slurry content
                                                    Of particular interest is a new design, with an        from the ash sump to a disposal
                                                    accuracy as low as 0.1% of reading when used           pond some kilometres distant
                                                    above 20% full scale. Such an instrument has           through a cement lined pipe which
                                                    greater intrinsic accuracy than the deadweight         rose vertically from the pump to
                                                    testers normally used for calibration (www.            run along a coal conveyor structure
                                                    crystal.com).


                                                                        vol 23 no 3 maintenance & asset management | May/June 2008 ME | 35
Experiences in Pump Condition Monitoring by Performance Analysis



      about 11 metres above the pump.                  To show the relative pump size, the          pressure measured across the orifice plate
      After a horizontal run of about 50           ash disposal pump design performance             read as zero! Investigation showed that
      metres, the pipe run took off at             with a 724 mm diameter impeller was 110          the pipe was not running full, and was
      right angles at a lower level, about 5       L/s @ 72m, with a 210 kW motor fitted.           in fact under symphonic vacuum. The
      metres above the pump, to the pond.          At the time of these tests, all pumps were       assembly was relocated to a vertical pipe
     The ash settled out in the disposal          of constant speed, and operated without          run but special temporary staging was
      pond and the carrying water                  discharge throttling.                            needed for installation and access to the
      overflowed to a reservoir. From                                                               pressure tapping points. Flow readings
      there, return water pumps supplied                                                            were now obtained, but the pressure drop
                                                   THE INVESTIGATION                                across the orifice plate was found to be 4m,
      a return water tank of uniform
      dimensions, located above ground,                 To keep each of the four plants             and thus restricted flow too much to be
      and therefore also above the ash             operating without town water make up             representative of the normal system. The
      sump.                                        being admitted to the system proved              assembly was removed and the normal pipe
                                                   difficult. The eventual spillage via an          section reinstated.
     Flow into the ash sump was irregular         effluent pond was to the main power
      in quantity, and also needed                 station cooling pond, such that chemical
      additional water to prevent ash              content built up to unacceptable levels.         SERENDIPITY STRIKES
      settling in the sump and ensured             Therefore, an investigation was begun to              Flow tests had just been run on the
      adequate net positive suction head.          fi nd the required flows from each pump to       return water pumps, such that the input
                                                   maintain a closed system operation. Pump         flow into the tank was constant and known.
     A level electrode in the ash
                                                   performance testing was required to provide      An ash disposal pump was started, and
      sump ensured that the level was
                                                   information that would lead to adjusting         the system set to auto operation. The tank
      maintained by opening an automatic
                                                   pump impeller sizes. The system was run          automatic outlet valve was controlled by the
      valve in the supply line from the
                                                   on water only for the tests. The same test       sump level electrode to open whenever the
      tank.
                                                   method was to be used for routine condition      sump level dropped to the low setting and,
     Similarly, a drop in tank level sensed       monitoring.                                      in turn, to close when the high level setting
      by an electrode in a side chamber                 Given the readily accessible tanks of       was reached. All flow out of the tank was
      started the return water pumps, or if        uniform dimensions from the top down for         replenishing the ash sump, according to the
      the inflow was insufficient, opened          most of their height, it was decided to use      detected sump level change and the auto-
      a town water supply valve for an             the rate of tank level change as the flow        valve. The level of the tank was measured at
      alternative source.                          meter. The concrete tank total capacity          regular intervals of one minute, and resulted
                                                   was 169 kL, with a volume of 27.75 kL per        in a plot of volume contained vs. time (see
     Return water and ash disposal
                                                   metre of depth in its parallel walls section.    Figure 1)
      pumps ran un-throttled.
                                                   The level change rate was measured with a             It was observed, when plotting the
     A small flow of water at high                stop watch and a weighted tape measure.          data afterwards, that because the level
      pressure was also supplied to seal           The flows of the return water pumps were         in the tank rose the return water pump
      the glands of the disposal pumps,            readily found at different tests by running      had a greater flow rate than the disposal
      and another supply injected into the         the system on water only. The input flow         pump. The gradient of the line showed
      sump at high pressure to maintain            was found to be 135.7 L/s. Pump suction          the difference. As tested, the return water
      the ash in suspension. These                 and discharge pressures were measured            inflow to the tank was at a rate of 136.7 L/s.
      flows were measured separately for           with test pressure gauges.                       From Figure 1 the gradient of water volume
      the plant investigation, but for a                The tank outlet automatic valve was         increase in the tank was calculated to be
      condition monitoring test on water           manually isolated for these test runs.           10.4 L/s. Thus, the disposal pump flow –
      could be shut off.                           Although the water level surged vigorously       clearly less than the incoming flow – was
                                                   as it gushed in, the level in the side chamber   136.7 – 10.4 = 125.3 L/s.
     This ‘tank-sump-pumps’ plant                  inside the tank was nicely damped and
arrangement (called ‘ash plant’ here, for          showed no oscillations.
brevity) was repeated four times, each such             Finding the output flow from the
arrangement serving a pair of generating           disposal pump was a different problem.           OTHER LEARNING
units. As the ash disposal pond was a              The initial method was to design and fit         EXPERIENCES FROM THIS
distance away from one end of the power            an orifice plate in the discharge pipeline.      SYSTEM
station, each set of pumps had a different         A section containing the orifice plate was           While not directly associated with
impeller size. The fi rst set had the largest as   made and inserted in a place where access        this main case study, other experiences
the conveying distance was further away.           was available from walkways adjacent to the      on this system were relevant to condition
Nearer to the pond, the ash plants had a           pipe. This was in the horizontal run some        monitoring and should be helpful to those
smaller impeller and motor. The last set was       11 metres above the pumps.                       new to pumps on any system.
larger, as revised site plans had a further             With the pump running, it came as
pond a greater distance away.                      a surprise to fi nd that the differential


36 | May/June 2008 ME | maintenance & asset management vol 23 no 3
                  Experiences in Pump Condition Monitoring by Performance Analysis


                                                   Tank level trend
                                                                                                                         much less, resulting in a pump flow much
                 130                                                                                                     greater than before, such that the pump
                                                                                                                         reacted by drawing more power. The power
                 120                                                                                                     draw exceeded the motor rating and the
                                                                                                                         pump tripped on high current protection.
                 110
                                                                                                                         Despite earlier assurances that the nozzle
                 100                                                                                                     had been recently checked, inspection
                                                                                                                         revealed that it had come off.
                  90
Relative level




                  80                                                                                                     The pump is not always to
                                                                                                                         blame
                  70                                                                                                          During the various investigations and
                                                                                                                         tests on the pumps described here the
                  60
                                                                                                                         operators reported that one of the return
                  50                                                                                                     water pumps was down in performance. The
                                                                                                                         investigating engineer went to the pump
                  40                                                                                                     house, which was unmanned and located
                                                                                                                         outside the power station and over 500m
                       1   2   3   4   5   6   7   8    9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
                                                                                                                         from the control room. On arriving, both
                                                            Time (minutes)                                               the pumps were observed to be running,
                                                                                                                         contrary to the operation agreed with the
                  Figure 1 Test plot of tank level change with time                                                      operator. A call to the operator confi rmed
                                                                                                                         that his panel showed only one pump to be
                                                                                                                         in service.
                  Wrong impeller fitted                                to service, but the annoying behaviour                  On a closer inspection the pump that
                                                                      continued. In the resulting investigation          was not in service was seen to be rotating
                       With a total of eight pumps of the             (which should have been done in the fi rst         in reverse. As most motor noise originated
                  same external appearance and dimensions,            place) the total head was measured using           from its cooling fan it appeared to be in
                  but with more than one interchangeable              the supplied tapping points in the pump            service. A very close look was needed to
                  impeller size available to suit the duty of         fl anges, and the flow was found using an          confi rm rotation direction. The suction
                  each location, care was needed so that the          expedient method. In this case, a double           and discharge isolating valves on these
                  correct impeller for each particular location       tip pitot tube was inserted through a gate         pumps are of the knife-gate type, operated
                  was fitted.                                         valve installed on the side of the 100 mm          by actuators. The limit switches on the
                        A pump to be sent for overhaul had its        diameter discharge pipe.                           actuators of the offending pump were out
                  impeller severely worn due to the abrasive               The resulting test points were consistent     of adjustment, such that the valves did not
                  nature of the ash slurry. A replacement             with the only available data, viz. the works       fully close. This allowed water from the
                  impeller was obtained from store and                test curve. They fell on an extension to the       service pump to re-circulate through its
                  fitted. After re-installation on site, operators    curve, as sketched in Figure 2. It was evident     partner, rotating it in reverse. This of course
                  reported that the pump could not maintain           that the system resistance had become              reduced the flow to the system.
                  the required flow. As the pump was newly
                  overhauled worn clearances were unlikely,              Head                                                 Normal operating point
                  so a head-flow test was run using the flow
                  measurement method described here.                                                                                                Test points
                  The pump performance was confi rmed as
                  below requirements and corresponded to
                  that expected from a smaller impeller. The
                  diameter was estimated from the head-flow                           Motor rating
                  curve, and proven correct when the pump                                                                              +
                  was dismantled.                                                                                                            +
                  System resistance reduction                                         Power absorbed by pump
                      A three-stage pump (duty 19 L/s @ 900
                  kPa) supplied high pressure water through
                  a nozzle to agitate ash in the sump so that
                  the ash was kept in suspension. The pump
                                                                                                             Flow
                  kept tripping on ‘high amps’ and it was sent        Figure 2 Test results showing that the system resistance had reduced, such that power absorbed
                  away for overhaul. The pump was returned            by the pump exceeded the motor rating.



                                                                                        vol 23 no 3 maintenance & asset management | May/June 2008 ME | 37
Experiences in Pump Condition Monitoring by Performance Analysis


                                                  Head                    Test point         Restricted
                                                                                              system                 Original system curve
System resistance increase
     The return water pumps, two nominally
identical (100 L/s @ 38m) in parallel,                                                                                       Combined curve – Pumps
pumped water in the ash system as                                                                                               1 and 2 together
described above. Normally one pump alone
was sufficient to handle the duty. Condition
monitoring tests on both pumps in new
                                                                                                                             Test points –
condition showed that they differed slightly                                                             Pump 2              pumps and system
in head-flow performance.
                                                                                        Pump 1                               in original condition
     After some service it was found that
both pumps were needed to run together
to maintain the same required flow. Pump
                                                                                               Flow
performance was suspected, but this time
a condition monitoring test was arranged         Figure 3 Sketch of test results where the system resistance had increased, leading to low flow to
                                                 the system
before any pump was removed for overhaul.
The flow was measured using the large tank
of regular dimensions as described before.               60
     The performance curve for both pumps
                                                         50
in parallel was drawn by adding the flow
points at selected constant head values,                                                                                                                           pump 1
                                                         40
with the result as sketched in Figure 3. The
test point fell on the curve expected for the    Head
                                                         30                                                                                                        pump 2
pumps in their original condition. However,
this point indicated that the system                     20
                                                                                                                                                                   System
resistance had increased (see the ‘Restricted
system’ curve). Inspection revealed that the             10
suction line, some 50m long, had suffered a                                                                                                                        Both
build-up of chemical deposition sufficient to               0
                                                                                                                                                                   pumps
reduce its effective diameter by a half. The                    0    50    100   150   200   250   300   350   400   450   500   550   600   650   700
pipe was buried and encased in concrete.                                                            Flow
Internal blasting would have been slow and
expensive so the economic solution was to
install a new suction pipe system.               Figure 4       Curves for two pumps showing performance in parallel and a typical system curve
     Note that in Figure 3 Pumps 1 and
2 are shown as having slightly different
head-flow characteristics. Given tolerances
in manufacture such differences do occur             With two or more pumps in parallel the                    performance problems in general, of which
between nominally identical pumps, even          actual system curve can be derived by test,                   many will be discovered and solved when
when new. From an operational viewpoint,         even if the static head (i.e. at zero flow) is                conducting condition monitoring tests.
this probably does not matter, but needs         not known. For a system with two pumps
to be taken into account for condition           in parallel and a known static head from
monitoring and plant investigations.             measurements on the plant, or from reliable
     These two latter cases show the             elevation drawings, then three points                         ACKNOWLEDGEMENTS
importance of head-flow measurements in          are available to plot the system curve (a                          The help of Kevin Speairs was much
detecting whether an observed performance        quadratic from the static head value).                        appreciated in running the tests and
shortfall is due to the pump, or its system,                                                                   retrieving data.
or both.
     All pump users should understand the        CONCLUSION
performance of pumps in parallel. Errors              A method has been described for                          REFERENCES
are often made in assuming an individual         fi nding the flow from a pump when access
                                                                                                               1.     Beebe R S, Predictive maintenance of pumps
pump’s duty when operating in parallel. In       to its suction chamber is restricted and no                          using condition monitoring,
Figure 4, with each pump having the same         other method is available. Given relevant                            Elsevier, UK 2004
head-flow performance each one would             circumstance the method is suitable for
be contributing half of the total flow and       plant investigations, and may also be
therefore be operating at 40 units of head,      applicable for condition monitoring. It is
and a flow of 207 units, not at twice the flow   hoped that the other lessons learnt will                        e author may be contacted at: Ray.Beebe@eng.
                                                                                                               monash.edu
of 330 when one pump is operating alone.         be helpful to people investigating pump


38 | May/June 2008 ME | maintenance & asset management vol 23 no 3