condition monitoring pays off for finnish pulp paper mill

8/14/2019 Condition Monitoring Pays Off for Finnish Pulp Paper Mill

1/16

CASE HISTORYPAYBACKPROFILE

Ulf SandbackaManager, Testing & InspectionsUPM [email protected]

Petri NohynekSales Engineer, Bently Nevada Asset Condition MonitoringGE [email protected]

Condition Monitoring Pays Off for Finnish Pulp & Paper MillHow UPM Wisaforest Uses System 1 Softwarefor Improved Asset Management

IntroductionThis article explores the use of System 1 software inconjunction with Trendmaster Pro data acquisitionhardware at a pulp and paper mill in Finland. Applied tonearly 50 discrete pieces of process machinery through-out the mill, the system has only been in operation sinceApril 2004, yet has already been instrumental in identify-ing and solving more than ten separate machineryproblems. An overview of the facility and the online con-dition monitoring system is presented, along with fivecase histories illustrating how equipment malfunctionsare being identified and resolved.

Background

UPM-Kymmenes Wisaforest facility is a pulp and papermill situated on the Gulf of Bothnias coast in Pietarsaari,Finland. The pulp mill was originally constructed in 1935and was designed around a sulfite process. The plantwas switched to a kraft process during a 1962 rebuild,and then in 1976 underwent major upgrades to thehardwood and softwood pulp lines. Production capacitytoday is 800,000 Air Dried tons per annum (ADt/a) of pulp and 180,000 ADt/a of kraft and sack papers. Themill generates its own power and is entirely self-sufficient in this respect. It utilizes two production linesthroughout, with exception of the new recovery island(WISA 800 REC project), discussed next.

The new recovery island at UPM-KymmenesWisaforest pulp and paper mill in Pietarsaari,Finland. The result of the WISA 800 REC project,

the new unit boosted the mills pulp output to800,000 ADt/a while decreasing environmentaldischarge to best-available-technology levels.


2/16

[Vol.25 No.2 2005] O R B I T 35


WISA 800 REC ProjectThe WISA 800 REC (RECovery) project had several goals,summarized in Table 1. The project entailed a single lineto replace the mills two previous recovery lines, whichwere in bad mechanical condition and date from 1962and 1975-1976, respectively. This new line is sized tomeet the mills pulp production capacity of 800,000ADt/a. To achieve this capacity, the two existing fibrelines were improved, resulting in better strength proper-ties of the pulp. Additionally, a new sawdust cooking linewas added, providing more flexibility in the raw materialbase by allowing the use of sawdust as feedstock for

selected pulp qualities, rather than burning it as hadbeen done previously. This delivers both environmentaladvantages (fewer emissions) and process advantagesby allowing the mill to draw from a larger pool of locallyavailable sources for pulp feedstock.

In addition to the goals for the WISA 800 REC project,the scope was likewise very extensive as summarized inTable 2. The main equipment suppliers for the recoveryisland were Andritz Corporation for much of the processequipment, Siemens AG for the main turbo-generator,

and Metso Automation for the process control system.For the condition monitoring (CM) system, GE Energy waschosen to supply their Bently Nevada solutions consist-ing of System 1 software, Trendmaster Pro hardware,and a 3500 monitoring system. Startup of the new recov-ery departments occurred on-schedule in April 2004.

Table 1

GOALS OF WISA 800 REC PROJECT

Become one of the most cost-efficient pulp millsin Europe

Increase pulp capacity to 800,000 ADt/a

Decrease the environmental discharge levels tobest-available-technology

Replace old and worn out equipment. Expand raw material options by using sawdust

for pulp production (instead of burning it)

Table 2

SCOPE OF WISA 800 REC PROJECT

Service Module Boffice building, control room, maintenance rooms

Evaporation Plant

7+ stages, (1,050 tons H 2O per hour, 82 - 85 % drysolids)

Recovery Boilersteam production 180 205 kg/s, 92 102 bar,492 505 C, 4,450 tons dry solids per day

Back-Pressure Turbine-Generator*143 MW

Causticising Plant10,000 m 3/day white liquor

Lime Kiln750 tons of CaO per day

Steam Condensate Treatment200 litres per second

Tall Oil Plant192 tons per day

* According to Siemens AG, this is the worlds largest back-pressure turbine generator used in the pulp & paper industry.


3/16


The Lime Kiln rotates at extremely slowspeeds (~10 rpm) on supporting rolls thatuse fluid-film bearings.

Table 3 CONDITION MONITORING SYSTEM SUMMARY

Department Numberof Machines Machine Types Monitoring System

Evaporation 6 pumps, mixers, fan Trendmaster Pro

Recausticizing 18 pumps, compressors, mixers, filters, fan Trendmaster Pro

Recovery Boiler 14 pumps, mixers, fans Trendmaster Pro

Lime Kiln 11 pumps, mixers, filters, fans, lime kiln, Trendmaster Prosupporting rolls and drives

Turbo-Generator 1 143 MW back-pressure steam turbine 3500 Series


4/16


5/16


was to determine project installation scope. It was decidedthat Wisaforest would be responsible for providing allLocal Area Network (LAN) connections along with locallyavailable power at each DSM location. The GE Energyteam would be responsible for installing all sensors,cabling, cabling shields, junction boxes, and DSMs. Inaddition, they would be responsible for the installationand configuration of the System 1 software along withintegration to the plants process control system.

When field hardware installation was approximately

20% complete, System 1 software installation com-menced. A significant element of the overall project wasto configure the software with appropriate monitoringparameters such as alarm settings, frequency bands,point labeling, and many other details. The machines inthe facility vary from one another in many ways includ-ing operating speed (1 3000 rpm), drive mechanism(direct, belt, and gear), and operating mode (constantspeed, variable speed, constant load, variable load).

This entails a high level of cooperation between numer-ous suppliers, plant personnel, and GE Energy todetermine and document the correct values for all set-tings, and then enter these values into the softwaresconfiguration screens.

In addition to the items already noted, a project of thismagnitude entails many other details, a few of whichare summarized below:

Server model and its installation location

Determination of LAN type (copper or fiber) andconnections thereto

Labeling of cables and sensors

System wiring topology

Interconnection of DSM hardware and 3500 System

to System 1

Ongoing dialog with machinery OEMs

External computer support


6/16

[Vol.25 No.2 2005] O R B I T 39


Coordination of machine test schedules with readinessof CM system, allowing analysis of start-up data

Integration with process control system and appro-priate data types for display to operators

The last bullet in the above list merits additional discus-sion. Originally, the project scope did not include aninterface between the process control system andSystem 1 software. However, as the usage scenarios of the CM system were further defined, the ability for oper-ators to view basic condition information using theirprocess control system was deemed important, whilestill providing in-depth diagnostic capabilities for rotatingmachinery engineers via System 1 softwares user inter-face. This resulted in two primary user interfaces: one foroperators, and one for machinery specialists. A bi-direc-tional OPC link was used for this interface, allowing theplant to not only import direct amplitudes and alarmsfrom the System 1 platform into the DCS, but also toexport numerous process variables from the DCS into theSystem 1 database.

[Editors Note: You can read more about the usefulness of process data correlation in a CM system and the importance of making select CM data viewable in the plant control system inthe article Best Practices for Asset Condition Management in theThird Quarter 2001 issue of ORBIT, pp. 46-47.]

Taking the New Recovery Unitinto OperationAs anyone who has been involved in a plant startup canattest, one of the most crucial times for the CM system is

when machines are tested and brought online for thefirst time. Problems that may not have been apparentat the factory may surface, or the installation of themachine may have introduced problems (e.g., alignment,piping strain, lube contamination, etc.). Consequently, avery important aspect of the project was to ensure that

the CM system was configured and ready for use aseach machine was started up.

The first test runs for the new recovery unit started on12 February 2004. Through careful advance planningand schedule coordination, the CM system was ready tobegin monitoring these machines. As other machineswere subsequently brought online, the System 1 configu-ration and commissioning were coordinated to coincidewith their start-up dates as well.

By 1 April 2004, when the new recovery unit officiallybegan full-time operation, all measurement points hadalready been collecting data for several weeks.Subsequently, the team turned its attention to finetuning alarm levels and other system configurationsettings, now that actual data was available from theoperating machines. During these adjustments, nomachine failures could be allowed, and faulty operatingconditions needed to remain visible. Wisaforest andGE Energy worked collaboratively to successfully accom-plish these objectives in a timely fashion.

As previously mentioned, although the plant started full-time operation on 1 April, machinery testing commencedseveral weeks prior to that date. During this start-upphase, a number of machinery problems were identifiedby the CM system, allowing proactive intervention andremedy even before the entire plant went live. Thisearly payback of the system and its usefulness duringstartup activities had been a high priority for theWisaforest project team and was part of the justification

for installing the system in the first place. All participantswere extremely pleased that the system demonstratedits value so early in the project. After full-time operationcommenced, the system continued to deliver value bylogging many other machinery saves. Several of thesesaves are summarized next.


7/16

40 O R B I T [Vol.25 No.2 2005]


Case History #1Problem: Bearing Lubrication

Machine: Secondary Air Fan

Unit: Recovery Boiler

The secondary air fan is a 600 kW direct-driven over-hung fan that is critical for the recovery boileroperation. Shortly after startup, abnormal changes intrends of the high-frequency data from the inboardbearing accelerometer were noted. Figure 1 is taken

directly from the System 1 software, showing a 2-month trend of high-frequency data from theaccelerometers on the inboard and outboard bearings.The elevated levels on the inboard bearing (blue) com-pared to the outboard bearing (orange) are readilyapparent.

12:2904SEP2004

12:2908SEP2004

12:2912SEP2004

12:2916SEP2004

12:2920SEP2004

12:24SEP200

25

20

15

10

A M P L I T U D E :

1 m

/ s 2

r m s /

d i v

Figure 1 Trend of high-frequency vibration amplitude from inboard bearing (blue) and outboard bearing(orange) showing elevated vibration levels at inboard bearing. The dips show the intermittent operation of the lubrication system, allowing the inboard bearing vibration to temporarily decrease to normal levels.

Access platform for Secondary Air Fan.


8/16

[Vol.25 No.2 2005] O R B I T 41


Spectral analysis suggested that the bearings outerring was wearing prematurely, and the root cause wasultimately traced to problems with the bearing lubrica-tion system. The prominent dips in the trend plotcorrespond to intermittent operation of the lubricationsystem, showing a marked decrease in vibration forthe inboard bearing when lubrication was flowingproperly.

Even though root cause was identified, implementingthe changes to the lubrication system was a lengthy

process, and the machine was required to operate inthe interim. Thus, although the bearing had to bereplaced twice during the first six months, the CMsystem proved very useful in scheduling thesereplacements, allowing the plant to monitor bearingdegradation closely and intervene at the right times,before catastrophic bearing failure and collateralmachine damage occurred. Also, these outages couldbe planned, allowing the bearing change-outs to beperformed when impact to production was minimized.

Case History #2

Problem: Resonance

Machine: Lime Kiln Driver

Unit: Lime Kiln

The lime kiln is a large machine, approximately 4.7meters (15.4 feet) in diameter and 135 meters (443feet) long, with extremely slow rotational speeds (aslow as 5 rpm). Two drivers provide rotational power,

and, depending on production conditions, the kilnmust run at different operating speeds. When the kilnran at higher speeds, higher vibration levels werenoted, occurring predominately at 2X. This led plantpersonnel to initially conclude it was an alignmentproblem, but realignment of the drivers did not correctthe situation and vibration levels remained elevated. Are-examination of the vibration data was conducted,this time by looking at phase and rpm data in additionto amplitude and frequency (Figure 2 and Table 4).

One of two drives for the plants massive Lime Kiln.

THE CM SYSTEMPROVED

VERY USEFULIN

SCHEDULING THESE

REPLACEMENTS, ALLOWING THE

PLANT TOMONITORBEARING DEGRADATION

CLOSELYAND INTERVENE AT

THE RIGHT TIMES.


9/16

42 O R B I T [Vol.25 No.2 2005]


07:1715AUG2004

07:1716AUG2004

07:1714AUG2004

07:1713AUG2004

07:1712AUG2004

07:1711AUG2004

07:1710AUG2004

07:1709AUG2004

09AUG200407:17

10AUG200407:17

11AUG200407:17

12AUG200407:17

13AUG200407:17

14AUG200407:17

15AUG200407:17

16AUG200407:17

TIME: 8 Hours/ div

360

270

180

90

0

20

25

0

5

10

15

P H A S E L A G :

1 5 d e g

/ d i v

A M P L I T U D E :

1 5 m m

/ s r m s

/ d i v

Figure 2 Amplitude/Phase/Time (APHT) plot of vibration data from lime kiln drive rollers with motor speedvarying between 991 rpm and 1083 rpm.

Amplitude reaches a maximum andphase undergoes a 90-degree shift(relative to phase at 991 rpm) atapproximately 1070 rpm. Vibrationoccurs at twice running (excitation)speed or 1070 x 2 = 2140 cpm (36 Hz).

Table 4

SUMMARY OF 2X AMPLITUDE AND PHASE DATA WITH RPM AND DATE/TIME STAMPS

Date/Time Speed Amplitude Phase Lag(rpm) (mm/s rms) (deg)

12 Aug 2004 10:29:30 991 1.50 51

12 Aug 2004 10:32:23 1030 3.72 64

12 Aug 2004 10:46:47 1031 3.76 62

12 Aug 2004 10:49:40 1033 3.81 6012 Aug 2004 11:06:58 1050 6.30 74

12 Aug 2004 12:59:24 1066 10.10 160

12 Aug 2004 13:05:10 1068 10.88 166

12 Aug 2004 13:13:49 1083 6.94 184

12 Aug 2004 13:54:11 996 1.82 52


10/16

[Vol.25 No.2 2005] O R B I T 43


Figure 2 is an Amplitude/Phase/Time (APHT) plotwhere the horizontal axis is time. It has characteristicsvery similar to a Bode plot, whose horizontal axis ismachine speed rather than time; namely, if themachine speed is changing markedly with time, anAPHT plot can show a resonance response, just as

a Bode plot. The classic features of resonance aretwo-fold: First, the filtered (1X, 2X, etc.) amplitude willincrease to a maximum at a rotational speed thatexcites the resonance, and then will decrease as themachine speed goes above this frequency. Second,the phase lag will undergo a 180-degree shift, gener-ally passing through approximately 90 degrees at thepoint of resonance.

While Figure 2 does not label each individual datapoint with its corresponding rpm, System 1 software is

capable of providing this information as a tabular out-put. The points clustered between 10AM and 2PM on12 August 2004 showed the most dramatic shifts inamplitude and phase, and coincided with a changeon the kiln from low-speed operation to high-speedoperation and back again. A tabular output of the datapoints in Figure 2 was generated, and a subset of thisdata is summarized in Table 4, clearly showing thecorrelation between amplitude/phase changes andrunning speed, and helping to confirm a structuralresonance at approximately 36 Hz.

During subsequent maintenance on the unit, the sup-ports for the drivers were stiffened and strengthened,raising the resonant frequency of the structure andeliminating the vibration problems.

[Editors Note: Resonance is a well-understood phenomenonin machines and structures, and operation of rotatingmachinery at a running speed that coincides with a reso-nance is never done deliberately. Sustained operation at a

structural resonance frequency can result in very high vibra-tion amplitudes, fatiguing connections and components, andprematurely wearing the entire machine. However, as thiscase history shows, it can be equally damaging to operatea machine at a speed that coincides with one-half of the res-onant frequency allowing the normally small 2X vibrationsgenerated by the machine to excite this resonance.Resolution of this problem was instrumental in ensuringthe kiln could achieve expected maintenance intervals andmaximum useful life.]

Case History #3

Problem: Bearing Failure

Machine: Exhaust Gas Fan

Unit: Recovery Boiler

The official dedication for the WISA 800 productionunit took place on 24 August 2004. The Prime Ministerof Finland was in attendance, and, as part of the cere-monies, pushed the maximum operation button,allowing the recovery boiler to reach world-record

production capacity for a time. This mode of operationrequired the exhaust gas fans to run faster, and the CMgroup began to notice an increase in 2X vibrationamplitudes on fan #3, as evident in the APHT plot of Figure 3.

Access platform for Exhaust Gas Fan #3.

THEPRIME MINISTEROF

FINLAND,PUSHED THE

MAXIMUM OPERATIONBUTTON.


11/16

44 O R B I T [Vol.25 No.2 2005]


14:3331AUG2004

14:334AUG2004

14:3317AUG2004

14:3310AUG2004

10AUG200414:33

17AUG200414:33

24AUG200414:33

31AUG200414:33

60

00

240

180

120

60

0.0

0.2

0.4

0.6

P H A S E L A G :

3 0 d e g

/ d i v

A M P L I T U D E :

0 . 5

m m

/ s r m s /

d i v

Figure 3 Amplitude/Phase/Time (APHT) plot of 2X vibration data from outboard bearing accelerometeron Exhaust Gas Fan #3. Note that amplitudes prior to the cursor location on the plot were so low that phasereadings would not trigger consistently; as vibration amplitudes increased, phase readings stabilized.

0 200 400 600

0

A C C O U P L E D

1 m

/ s 2 / d i v

50 ms/div

Figure 4 Unfiltered timebase plot from outboard bearing accelerometer on Exhaust Gas Fan #3.Note characteristic ringing phenomena as inner race defect is impacted by rolling elements.


12/16

[Vol.25 No.2 2005] O R B I T 45


Closer examination revealed that the outboard bear-ing of fan #3 had likely sustained a crack in the innerring, which can be noted in the timebase of Figure 4.The characteristic ringing phenomenon observablein the timebase occurs at the inner ring defect fre-quency as the cracked inner ring repetitively rotatesthrough the load zone and the rolling elements contactit with greatest force.

Subsequent to that event, the bearing has been moni-

tored closely, allowing operations to continue withoutreplacing the bearing. The defect does not appear tobe progressing and is not serious enough to necessi-tate a bearing replacement until a more convenienttime can be scheduled.

[Editors Note: It is not clear whether an invoice for a newbearing will be sent to the Finnish Prime Ministers office. ]

Case History #4

Problem: Faulty Coupling

Machine: White Liquor Pump

Unit: Recausticizing

A typical pulp mill has hundreds of pumps. AtWisaforest, 15 of these were deemed suitablyimportant to connect to the CM system. In late May2004, as shown in Figure 5, increased vibration levelswere noted on the motor driving the white liquorpump. Further analysis revealed that the rubberelement in the coupling had deteriorated, allowingmetal-to-metal impacting. The coupling was subse-quently repaired and vibration levels returned tonormal.

White Liquor Pump showing motor and coupling guard.

FOR WISAFOREST,

THE ECONOMIC BENEFITS

HAVE SUBSTANTIALLY

EXCEEDED

EXPECTATIONSRESULTING IN AN

ESTIMATEDPAYBACK TIME

FOR THEIR INVESTMENTOF

JUST 8 WEEKS.


13/16

46 O R B I T [Vol.25 No.2 2005]


Case History #5

Problem: Bearing Deterioration

Machine: Mixer Adjacent to Rotary Filter

Unit: Recausticizing

Rotary filters are one of the most difficult machines tomonitor since their rotational speed can be extremelylow as little as 0.5 rpm. The bearings are fitted withaccelerometers and acceleration enveloping is one of

the signal processing techniques used to help identifydegradation and other anomalies.

In early September 2004, the CM group began tonotice increased vibration levels on the filters inboardbearing, observable in both the enveloped amplitudeand the high-frequency amplitude trends (Figure 6).

14:3315JUN2004

14:3322JUN2004

14:3308JUN2004

TIME: 24 Hours/div

14:3301JUN2004

14:3327MAY2004

14:3318MAY2004

18MAY200414:33

25MAY200414:33

01JUN200414:33

08JUN200414:33

15JUN200414:33

22JUN200414:33

360300240180120600

0.0

0.5

1.0

P H A S E L A G :

3 0 d e g

/ d i v

A M P L I T U D E :

0 . 1

m m / s r m s /

d i v

Figure 5 Amplitude/Phase/Time plot of 2X data from accelerometer on motor driving the white liquor pump.Note abrupt increase in vibration amplitude beginning on 30 May 2004. A deteriorated coupling insert was foundto be the cause, and was replaced on 1 June 2004, returning vibration levels to normal values.

The rotary filter. Accelerometers were mounted on the filterbearings and the gearbox (green, upper right); however, theproblem was traced to an unmonitored mixer underneaththe filter (concealed in lower left corner of photo).


14/16

[Vol.25 No.2 2005] O R B I T 47


13:47

08OCT2004

13:47

10SEP2004

13:47

13AUG2004

13:47

16JUL2004

13:47

18JUN2004

13:47

21MAY2004

13:47

23APR2004TIME: 7 Days/ div

1.2

0.8

0.6

1.0

0.4

0.2

0.0

A M P L I T U D E :

0 . 0

5 E n v m

/ s 2

p

k / d i v

13:4708OCT2004

13:4710SEP2004

13:4713AUG2004

13:4716JUL2004

13:4718JUN2004

13:4721MAY2004

13:4723APR2004

TIME: 7 Days/ div

1.5

1.0

0.5

0.0

A M P L I T U D E :

0 . 1

m / s 2

p

k / d i v

Figure 6 Amplitude trends from rotary filter bearing accelerometer showing high-frequency acceleration(top) and enveloped acceleration (bottom). Note increased amplitudes in both signals beginning onapproximately 3 September 2004.


15/16

48 O R B I T [Vol.25 No.2 2005]


The enveloped acceleration timebase (Figure 7)showed very clear evidence of periodic impacting, butexamination of the spectral components did not yieldany frequencies corresponding to the bearing geome-tries or rotative speeds of the filter or its gearbox. Avisual examination of the filter gave the reason: it wasnot the filter at all, but rather a separate mixer, locatedbelow the filter, with a damaged bearing. Although themixer was a totally unmonitored machine , the impactvibrations occurring from its faulty bearing were beingmechanically coupled into the accelerometer on thefilter bearing, located nearby. The root cause wasfound to be broken lubrication piping feeding the mixerbearing, which was subsequently repaired. However,the bearing had been irreparably damaged and had tobe replaced. This case history is particularly notewor-

thy in that it demonstrated the sensitivity of the moni-toring system to detect changes in even unmonitoredmachinery. While certainly not recommended as adeliberate CM strategy, it was an unexpected and,as it turned out valuable fringe benefit.

PaybackWhile the users of the CM system are very pleased withits diagnostic capabilities, it is important that we are ableto continually justify to plant management and opera-tions the value of condition monitoring and diagnosticsin general. For Wisaforest, this translates to economicbenefits, and those benefits have substantially exceededexpectations. Several of the machines highlighted inthese case histories have a critical role: they will causea complete stop in plant production if they do not run,

0 2010 30

1

0

1

A C C O U P L E D

0 . 2

E n v m

/ s 2 / d i v

2k ms/div

-0.02+3135.94 ms0.32 hertz

Figure 7 Timebase plot of enveloped acceleration showing clear evidence of impacting. Spectral analysisyielded frequencies that did not coincide with bearings used on the rotary filter, leading the plant to look forfaults elsewhere.


16/16


representing substantial lost production costs. Plant per-sonnel have avoided several total outages through use of the system, resulting in an estimated payback time fortheir investment of just 8 weeks.

Both the maintenance and production departments nowview the system as far more than just a tool for strength-

ening preventive maintenance capabilities it is viewedas a tool for increasing the mill productivity. Strong credi-bility has been established with management regardingthe value of condition monitoring and its role in ensuringplant uptime, leading the plant to consider expanding thesystem to additional equipment.

SummaryWisaforest is achieving ongoing success with their CMsystem for several reasons:

Proven, quality technology from a knowledgeablesupplier was chosen as the basis for the plantsCM program.

The plant enlisted the assistance of the supplier tohelp install and implement the technology correctly.

Adequate transducers were installed where feasible,including speed/phase, rather than just vibration,

allowing confirmation of faults that would have beendifficult or impossible to isolate when limited to onlyamplitude and frequency data (e.g., case history #2).

Start-up activities were coordinated to include thecondition monitoring system as must have capabili-ties before a machine was brought online.

Plant management made certain that everyoneunderstood the CM programs goals and objectives,and that there was buy-in from all parties. This helpedensure that the system would be used proactively andconsistently.

The CM system was integrated with the process con-trol system, allowing operators to have early visibilityto developing conditions, and allowing process data tobe available for correlation with vibration data whenperforming in-depth diagnostics.

Results were documented, allowing the users to quan-tify the systems value to management and otherstakeholders in the plant.

Consequently, the WISA 800 REC project has led to notonly a world-class facility, but world-class asset man-agement practices and world-class results.

STRONG CREDIBILITYHAS BEENESTABLISHED WITH

MANAGEMENTREGARDING THEVALUE OF

CONDITION MONITORING.

condition monitoring pays off for finnish pulp paper mill

Documents