Weird and Wonderful Pump Failures

Or:It was not so wonderful at the time but I did learn something.

Topics1. Reactor Feed Pumps2. Wash Water Pumps3. Amine Pumps4. Fertilizer Plant pump5. Sulphur Transfer Pump6. Stripper Sour Water Pump7. Low Flow Recirculation8. Foreign Objects

1: Reactor Feed Pumpso Three 100% pumps, two are steam drive, one is electric drive.o Typically the steam drives are running (1 per train) with

electric drive on stand-by. o Pumps are 8 stage 3” IJ’s, BB5 designo Single bellows style seal with SC rotating face, TC stationary

face,o Plan 32 seal flush (66 °C gas oil) and Plan 62 steam quencho Tilt pad axial and radial bearings, full vibration probe suite

14.6 ft

Reactor Feed Pumpso 1304 KW at 5600 rpmo Rated Flow: 140 m3/hr (616 gpm)o Rated Head: 2201 m (7221 ft)o Discharge Pressure: 20,021 kPag (2903 psig)o Normal temperature: 232 C (450 F)o Max. Temperature: 288 C (550 F)o NPSHa: 13 m (42 ft) o NPSHr: 11 m (36 ft)o Material class is API A-8 (316 AUS) but case is carbon steel

with 316L cladding

Reactor Feed Pumps

0

500

1000

1500

2000

2500

3000

0 20 40 60 80 100 120 140 160 180 200FLow (m3/hr)

HEA

D a

nd P

WR

HEAD (m) PWR (kW)

NOTE FLAT SLOPE

Min. Flow

Rated Point

Start-up and Operation issues:o When units are started for the first time (at minimum flow)

the vibration levels and performance are acceptable.o But after being in service for a couple of weeks one of the

units goes down on high vibration.o Teardown shows rubbing damage to wear-rings and balance

bushing, some loose case wear rings, and movement of the 7’th stage impeller away from suction then back towards suction into the diffuser ring.

Back of 7’th stage impeller and 7’th stage diffuser

Start-up and Operation issues:o Initial failure is thought to be due to breakdown of the wear-

ring material and internal rubs caused by poor warm-up procedures, hot misalignment, and cavitation.

o This leads to a great deal of work doing hot alignment and base deflection measurements with lasers, baseplate grouting repairs, warm-up piping and procedures mod’s etc.

o The initial failed unit is re-built with the wear ring coatings applied using a different process and the Operations staff are told to monitor/control the warm-up process better.

Start-up and Operation issues:o There are more failures of a similar nature…..even on the first

unit with the repaired base-pad, the tighter warm-up controls and more accurate hot alignment.

o Other issues with the units are discovered n Seal failures due to poor balancing of seal flush rates to

each end of the machine…the piping gets changed.n More insulation is added to the casing, balance line loop

and warm-up piping to prevent blockage.n Modifications to warm-up piping are proposed to improve

flow control when units are cold but never implemented.

Start-up and Operation issues:n Seal face damage due to wet quench steam flashing inside

the seal. Additional steam traps are added.n The anti-rotation keys at each case ring are increased in

length to positively clamp the rings in place axially. n Lube oil water contamination caused by steam leaking

from the turbine glands (2-3 gallons/day) fixed by installing In-Pro seals on the bearing housings.

o After a few more failures, sometimes 7’th stage, sometimes 8’th stage we finally started to zero-in on the main failure cause…..

Start-up and Operation issues:o The minimum flow control system

n During normal operation the closed minimum flow valve would cool off and not open properly when forward flow to the process was cut back at the fired heater.

n The small filters in the pump discharge flow meter purge oil system would continually plug up taking the flow meter off-line or make the measurement in-accurate. The operators had to manually control the min-flow valve with absolutely no information as to the actual pump flow rate.

n The pump discharge pressure at less than min-flow was enough to force the min-flow valve closed. The actuator was not sized big enough to keep the valve open!

Summary o We got caught up believing the drop in flow was caused by

damage in the pump when it was actually the other way round….the drop in flow (dead-heading) was causing the pump damage.

o When the min-flow control system was fixed the pump reliability went way up.

o Keep going back to the physical evidence….the backward moving impellers should have focused us to a hydraulic cause.

2: Wash Water Pumpso Three 100% pumps, One for each train, one spare. o Electric drive with gearbox reduction to 203 rpmo Pumps are 3 3/8 X 6 VTEM vertical triplex recipso Class S-1 design, sour specification for fluid endo Sleeve bearings, forged crank, aux lube oil system o Two amine pumps are similar design but different operating

speed.

Wash Water Pumpso 300 hp at 203 rpmo Rated Flow: 30.5 m3/hr (616 gpm)o Suction Pressure: 340 kPag (49 psi)o Discharge Pressure: 18,453 kPag (2676 psig)o Normal temperature: 95 C (203 F)o Max. Temperature: 98 C (208 F)o NPSHa: 27 m (88 ft) o NPSHr: 15 m (49 ft)

Wash Water Pumpso Numerous issues from day 1

n Packing leaks due to poor plunger and push rod alignment

n Poor valve lifen Broken crankshafts due to

fatigue initiating at hot-short tears in forging

n Cracked fluid head allowing water into crankcase

Wash Water Pumpso Failure of forged pressure head

Wash Water Pumpso Failure of forged pressure head

Wash Water Pumpso Major review conducted of operating history and

head design including FEA by OEMo Fortunately we had lots of pressure data due to valve

performance and cavitation studies.

Wash Water Pumpso Results of studies show that pump should be

safe from fatigue failures even at off-spec operating conditions.

o So why did the block fail? o HINT---The FEA analysis was based on the

OEM’s drawings for the head.

Wash Water Pumpso Dimensional checks of the failed head showed that the radius

at the bottom of the suction valve pocket was smaller than design by 1/8 inch!

Wash Water Pumpso Using the actual radius in the FEA model indicated

that fatigue was possible.o Suspect that the machinist used the wrong size tool

when the suction valve pockets were being machinedo All similar heads on site checked… one other head

had to be replaced.o Considered shot-peening the radius area to improve

fatigue resistance but no further failures occurred.

3: Amine Pumps

Amine Pumpso Lean Amine API Type BB3 pumps protected by three

shutdown systems:n Low feed drum leveln Low discharge pressure n High vibration

o Nuisance trips cause changes to controls:n The low discharge pressure SD is removed for start-up but

not reinstated after start-upn The low level trip on the feed drum is tied to the pump

shutdown.n High vibration SD’s changed to 2 out of 2 votingn Vibration SD set points raised

o No one caught that the vibration levels were set too high.

Amine Pumpso Residence time in the feed drum is very shorto Low level trip was lowered to 2% from 15% to prevent

nuisance pump shutdownso Level transmitter sensing legs were equalized with glycolo The glycol in the sensing legs became unbalanced, and

indicated a higher feed drum level than what there actually was and the pump ran dry

o Another incident elsewhere in the plant took the operators attention away from operating the plant to shutting the plant down

Amine Pumpso Outside operator attempted to shut down the operating pump

because it was “making noise”o The control system auto-starts the second pump and shortly

thereafter it starts making a loud noise too.o Vibration instruments did not shut the pumps down.o “Pinning out” the local start button successfully stopped these

pumps.o Remarkably, both pumps were still capable of moving liquid

when drum level was restored.

First rotor, 1’st stage impeller warped and cracked

First rotor, 2’nd stage impeller thrust damage

First rotor, general condition of the case

Second rotor, 2’nd stage impeller thrust damage, broken wear rings

Second case, bottom half wear ring land damage

Second rotor, impeller keyway damage

Amine Pumpso Removal and repairs began within 2 days.o Pumps were repaired in tandem utilizing two repair shops,

both shops working 24 hours per day.o Extensive weld repairs needed for cases and impellerso New shafts, wear rings, and bearings installedo A temporary 14 stage rental pump was installed in case there

was an issue with any of the repaired pumps. Skid pump was never operated.

o Pump repair was 16 days stream to streamo Rental pump installation was completed in 14 days.AN AMAZING AMOUNT OF WORK IN A SHORT TIME

Rental pump installation

Rental pump installation is ready for service

4: Fertilizer Planto 1,500 hp double suction pump running at

3,600 rpm. Double voluteo Pump design used extensively in Europe with

very good reliability in similar serviceo Original impeller is 6 vane designo Impeller and/or shaft failure approx every 6

months

Impeller Damage

Fertilizer Planto Change to 6 staggered vanes and very large radius on keyway

under impeller ….increased life to approx 4 yearso Further change to 7 vane staggered and profiled case

cutwaters …. problem has basically gone away.o Thoughts on the service/pump:

n very high energy pumpn 6 vanes and double cutwater, a common European design,

often leads to very large pressure pulsationsn pump initially designed and checked for 50 Hz power

(3000 rpm) but not for 60 Hz (3600 rpm)

5: Sulphur Transfer Pumpo Two vertical centrifugal VS4 style pumps installed in vessel.o “A” pump top bearing deterioration detected by periodic

vibration monitoringo Switched to stand-by “B” pump but very little flow

developed. “B” unit had not been operated for 12 months

Sulphur Transfer Pumpo Pulled “B” pump and found extensive blockage due to chunks

of 3/8 to 3/4 inch sulphurcrete in the impeller and suction.o The original strainer was designed with maximum perforation

of 1”.

Sulphur Transfer Pumpo Raise the pump suction 3 feet by

shortening the column o increase the normal liquid level in

vessel to 20% to maintain NPSHaat required value.

o pump strainer re-designed to block particles larger than 3/8”

o Operating procedure put in place to switch pumps on a fixed schedule

6: Stripper Sour Water Pumpo Corrosion found on 4140 shaft near stuffing box and

on case wear ring seats of the A216 casing halves

Stripper Sour Water Pumpo Corrosion mechanism is probably related to

selenium concentration…not normally expected in this process.

o Suspect that upstream oil sand feed is responsible o Upgraded the shaft components as follows:n – Base material to 12Cr.n – HVOF coated

7: Low Flow Re-circulationo Ansi 1 X 2 -10 condensate pump o 2 years of operation at high head – low flow o 2 thrust bearing failures in 2 yearso Pulled impeller and found ¼ inch metal loss

due to improper adjustment o Strong evidence of discharge recirculation

Low Flow Recirculation

Low Flow Recirculation

8: Foreign objects

Foreign objects

Commento Some say failure analysis is easier now because of the sophisticated data

collection and trending tools we have.o To this I say ……

Commento Some say it is failure analysis is easier now because of the sophisticated data

collection and trending tools we have.o To this I say ……

My thanks to:o Chris Gilmouro C. Dale Alleyneo Brad Hatton o Scott Masterson o Matt Pickardo Hector J. Amaya

o Jim Dziadyko Frank DiRoccoo John Hogano Gordon Lawrenceo David Goddard

for providing pictures and information for this presentation!