You are on page 1of 8

A Successful Preventive Maintenance Program

for an Electric Power Plant


Katsushige Onodera - Hitachi Ltd. - Hitachi
Yoomei Katoo
Hitachi Ltd. Hitachi
Key Words:

Preventive Maintenance, Electric Power Plant, Expert System, Maintenance Planning.


Diagnostics, Life Prediction, Life Extension, Turbine, Generator, Crack Detection,
lnsulat ion Deterioration Detect ion.

SLMXARY & CONCLUSIONS

1 . INTRODUCTION

The electric power capacity of thermal power and


nuclear power plants in Japan are increasing at a
rate of 2.7%/year.
The operating conditions of those plants are
severe, with long, continuous operation for
power 'Iants
and dai
start-stop
Operat ion for
power plants.
These
operating conditions directly affect the equipment
life and the system
Of the power
plants.
Continuous
Or
start-stop
operat ion, without fai lure needs an excel lent
plant preventive maintenance planning and
diagnostic system for each equipment and component.
This paper presents a comprehensive preventive
maintenance program for electric power plants with
an objective of "Prevention of plant forced
outage".
forced Outages
shutdowns) occur with failure of critical
equipment and systems during plant operation,
which include both the primary equipment as well
as auxi 1 iary equipment.
As a result, we have developed a diagnostic
system, failure detect ion methods, and 1 ife
estimation systems for critical equipment.
Good results have been obtained, with a decrease

Today's generating electric power capacity of


thermal power plants and nuclear power plants in
Japan is about 132 GlJe (78% of total power) and
s t i l lincreasingat a rate of 2.7%/year.
In recent years, nuclear power plants require
continuous operat ion, and thermal power plants are
operated with load-fol lowing or daily start-stop
operation for adjusting the level of electric
power generated. These operating conditions have a
direct effect on equipment life and system
reliability because of the thermal cycle stresses
of load change or long cant inuous operation.
Either reliablecontinuous operation, o r daily
start-stop operation, can be accomplished with
excellent plant preventivemaintenance.

preventive
This paper presents a successful
maintenance program which can plan preventive
maintenance (pM) using diagnostic techniques of
remaining life for each equipment.
The need for development of a comprehensive
preventive maintenance program is shown in Fig. 1

(Ref.

l).

PLANYIhlG SYSTEM

plants and nuclear power plants.

The comprehensive planning system has been


developed to assist in implementing a preventive
maintenance program.

0149-144)(/93/$3.00
8 1993 IEEE

146

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

Failure
Rate

(XI

\IL

/'

Bilm-in
ReZion

'
e
/,
S

Useful L i f e
Reqion

Year-ocr
Repion

Ciaqnosis Data o f Renainirg L i f e


Operation B maintenance Data

Recomecdat ion o f izsoect ion. i a o r a v e ~ e f i r

R e l i a b l e 0pera:ion

This system is composed of data inputs on


devices used in the plant, the data base of each
equipment, a common data base, and the maintenance
planning and recommendation system.
The structure of this system is shown in Fig. 2.

I
I

Plants

Fig. 1 The Need f o r Development of Comprehensive


Preventive Maintenance Program

This system can identify the following items


from the manufacturer (Hitachi Ltd.) to the using
ut i 1 i ty.
(1) Recommendat ions for improvement, or repair
items, for critical equipment and systems from the
standpoint of high availability and life extension.
(2) Recommendations of inspect ion criteria and
items for critical equipment and components.

(3) Recommendat ions for pending quest ions and


remaining matters
related
to
replacement,
inspect ion of equipment, and components.

Actual maintenance, operat ion, and failure data


of each power plant is inputted to the data center.
In particular, the maintenance data includes
diagnostic data of deterioration and the remaining
1 ife of equipment in the power plants. Naintenance
data are then analyzed f o r trends of deterioration,
remaining life o r inspection results.
This data is stored in the common data base o r
in the equipment data base.
The common data base is used for storage of common
data such as plant operat ion records, maintenance
schedule and plant shutdown failure data.
The equipment data base stores inherent data of
equipment such as diagnostic results, maintenance
records, and trends of remaining life.
The maintenance planning and recommendation
system can prepare the recommendations f o r
required maintemance items at each plant,
systematically, without omission, and in a timely
manner.

'

(Plant]
I RK P/an:Short/Long
I

Team

Maintenance Work

. Detection Crack and Diagnosis


. Nondestructive Test (NDT)
. Calibration of Instrument etc.

Diagnosis o f Deterioration
and Remainina L i f e
I

I Maintenance

-LJ-

Records

Preventive Maintenance Plan


. Inspection Plan
. Improvement Plan
. Recommendat ion Items

I
-

Equipment Data Base


Maintenance Records
Specification of Equipment
Failure Data
Trend of Remaining Life
-CY
I

1 Actual

Common Data Base


Plant Specitication
Operation Time
Start/Stop Times
Maintenance Schedule
Operation Plan

Data from Each Plant


Maintenance Records
Operation Records

Fig. 2 The Structure of Hi tachi Comprehensive Preventive Maintenance


Planning System
1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

147

3. REM INING LIFE DIAGNOSTIC SYSTEhIS

3.1.1

Electric resistance method

I t is known that metallurgical changes and creep

We have developed the diagnostic systems for


estimation of remaining life for both primary
equipment and auxi I iary equipment from the
standpoint of preventing plant forced outage.
The diagnostic systems of remaining 1 ife for
equipment are shown in Table 1.
By way of example, the diagnostic system f o r the
turbine and generator wi 11 be explained.

damage accumulate during long-term use under high


temperature, where carbide compound prec ip i tates
and
grows near
grain boundaries.
This
precipitation
gradually
decreases
electric
resistance when creep damage occurs. The
electrical resistance method uti I izes the
correlation between electric resistivity and creep
damage.

3.1.2
Table 1 Diagnostic Systems of Remaining Life
for Equipment
Equipnent

Turbine

Generator
Motor
h p

Conponent

Diagnostic Het hod


Electric resistance test
Hardness test, Replica method
Replica method
Hard etching

Rotor.
Blade.
Cas i ny.
Main Stop
Va I ve

Stator Coil
Shaft

-+ CREEP
FAT I GL!

3.1.3 Observation of microstructure

H-Y Yap method

D Map nethod

sys lem

X-ray diffraction method


Optical micro-crack detection method
On-condition monitoring - .lh.vibration

Fan
Electric

Va I ve
Packing

actual operation data

Beat
Exchanger

Tube

Eddy current test for tube thickness


Data analyzer

Pipe

Tube

Ultrasonic test (bl) for tube thickness


Trend of thickness with analyzer
(CAT)

Control

carbon seal
0 ring

Hardness measurement

In order to increase the accuracy of evaluating


the remaining life, it is necessary to obtain a
correlation between hardness deterioration and
creep damage. This relationship can be derived
from the test of the same, or similar material as
the original one.

Trend of cooling water teqperature


Comparison between life test data and
actual operation data

The replica method duplicates a condition of


grain boundary on rep1 ica film. The microstructure
is observed with the replica method and creep
damage is evaluated qual itat ively.
For heat resisting alloy steel, deterioration
has shown by carbide compound precipitates and
growth near grain boundaries. When
the
deterioration continues, voids appear on a track
where the carbide compound exists.

3.1.4

Temper embrittlement

Two methods for measuring the amount of temper


embrittlement have been developed for the actual
diagnoses. One diagnosis is the grain boundary
etching method. The degree of temper embrittlement
is measured with fracture appearance transit
3.1 Diagnostic system of remaining life
temperature (FATT) and proportional to the length
for turbine
and depth of pits in the grain boundary after hard
The steam turbines are operated under severe etching of the material surface.
We have
operating conditions, with high temperature, high developed this temper embrittlement evaluation
pressure, daily start-stop and load following technique and it is ready for field use.
operation. The material of the turbine, under
Another way to measure temper embrittlement is
these conditions will deteriorate as a result of
damage
from creep, fatigue and
temper the small punch (SP) test method by using metal
embrittlement. In the worst case, damage will lead sampling and a mechanical test. This test is not a
t o destruct ion. Therefore, the inspect ion and nondestructive test, but only a-small sample (lox
diagnosis f o r estimating the remaining life of the 1OXO.5 mm in size) is used. It is possible to
remove a sample from an actual component.
turbine is important (Ref. 2, 3).
The diagnostic process of determining the
remaining life for a turbine rotor is shown in Fig. Actually, it is not easy to determine the degree
of material damage, because the factors of fatigue,
3.
creep, and temper embrittlement are complex.
Rod Drive
Mechanism

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

-~

Uestructive

Electric
Res istance
Test

Hardness
Test

Sampling
Test
Test

Test
I

Potent i a 1
Difference
Decrease
in Electric
es is tance

Decrease
in
Hardness

E l
Temp.

FATT

SP Energy

Maintenance Records

Fig. 3 The Diagnostic Process of Remaining Life Evaluation for Turbine Rotor

In order to raise the accuracy of diagnosis, mechanical,


and
thermo-mechanical
cycling
various techniques are implemented in combination. resulting from the start and stop operat ion.
Moreover, a1 1 the procedures are systematized on a
The Insulation deterioration mechanisms and
computer-aided basis. Some of procedures use their detection method are shown in Fig. 4.
Expert Systems using a knowledge data base.
\\%en new, insulation composites of mica-resin
are a voidless structure. Voids in the insulation
3.1.5 Application f o r Turbines
The diagnostic systems for evaluating the are gradually produced by deterioration. Some
remaining life for turbines have been applied to voids become large and combine with others. The
50 thermal power plants, and 250 equipment during dielectric strength of mica resin then decreases
seven years. As the result of diagnoses, six to a critical
turbine rotors, five turbine casings,
and eleven
we have developed two new remaining life
main turbine valves
were replaced. Equipment which
estimation methods and their expert system of
were evaluated and found to have adequate insulation diagnosis (Ref. 4, 5). These are
remaining life will be diagnosed again in a four- discussed be low.
year period.
3.2. 1 N-Y Map method
One estimation is the "N-Y map" method that
3*2 Remainingi f e estimationsystem of determines how much insulation deterioration is
stator coil in generator
caused due to various factors affected by the
Insulation deterioration Of the Stator Coils in nuber of start-stops (y) and the operating time
large generators is caused by electrical,
( Y ) . The data is taken from the generator
operating history.
1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

149

Part i a 1 d i scharge
test ; Maximum

Fig. 4

Insulation Deterioration Mechanisms and its Detection Nethod


for Stator Coil in Generator

The main factors influencing insulation


deter iorat ion are the repetition of thermomechanical forces due to starts and stops of the
power plant, thermal aging, electrical stress
aging, and mechanical fatigue due to operation of
the plant.
Each deterioration ratio of the
insulation strength caused by the above-ment ioned
factors can be obtained from many model aging
tests.
Then, residual breakdown voltage can be
obtained by the product of the deterioration
ratios, as shown in eq. (1).

16
14

162217 hr

=5611

Tiaa
VR = 73.54 I

12

A 10

Stw
U

tb!%

a
6
4

2
00
Klm

Vr: residual BDV (Breakdown Voltage)


Vo: initial BDV
Ve: residual BDV due to electrical
deterioration
Vt: residual BDV due to thermal deterioration
Vh: residual BDV due to heat cycle deterioration

q*lrpticil hours (Hr) x10o0O0

Fig.5 Insulation Strength Estimation


Curve by N-Y Map Method

The field data was obtained by examining the


generator coi Is actually used, after long
operat ion. The measured residual breakdom voltage
The insulation strength estimation curves of the (BDV) and estimated BDV of the N-Y map aere
N-Y map are shown in Fig. 5 (Example of display compared. Both of them were nearly equal.
Comparison between the est hated BDV using the N-Y
fram expert system).
map and the measured BDV is shown in Fig. 6.

150

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

0 : Average estimated value


A : Minimum estimated value

0 :Average measured value

A :Minimum measured value

Measured
BDF

(3

Ol
Fig. 6

IA
IB
IC
ID
IE
IF
Estimated BDV of Actual Used Generators

Comparison between Estimated BDV by Xi-Y Map


and Measured BDV

3.2.2

D Map method
The another estimation technique is the "D map"
method. This method evaluates the voids that aere
produced and measures discharge electrical Fig. 7 D Map
magnitude by nondestructive tests, such as tan 6
and partial discharge tests.
Values that are useful in detecting void
discharge i n the electrical diagnoses, are AC
lzn
current change ratio ( A I ) and tan 6 change ratio
( A 2 ) , which represent the extent of average Sln0
deterioration, and &ax, which represents the
extent of local deterioration. The relationship >
between
the
values
was
obtained
from 2''
nondestructive tests, and BDV has been studied by
40
examining many used coils and is obtained using ;
a
:
10
the following eq. (2) :
r
al

VS=lOO

-l.s(A -0.8)

-27.4 log "ax / ( 1.5X 10E-9)


where
Vs: residual breakdown voltage (%)
A : discharge parameter (=AItA2)
Qmax: maximum discharge quantity (pC)

Estimated value:Vs(%)

(2)

Fig.
8 Comparison between Estimated BDV by N-Y
Map and Measured BDV

residual I ifetime curve

The D map is shown in Fig. 7 (example of display


from expert system).
The comparison between estimated BIN by the D map
method and measured BDV is shown in Fig. 8.

.....

_ _ N-Y

Map mean value

Map min. value


-N-Y
D Map mean value

-D Map Min.

value

By using the knowledge base of residual life


estimation, residual breakdown voltage is finally
decided by comparison with the magnitude of the
residual breakdown voltages deduced from the K-Y
map and the D-map.
Residual lifetime can be obtained from the
difference between inspection data and the
0
present
Operation
Years
estimated lifetime, at which time the residual
breakdown voltage will become 40% of the initial
value in the life estimation curve (Fig. 9).
Fig. 9 Residual Lifetime Estimation Curve

-+

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

~~

151

3.2.3

E.upert system for insulation


diagnosis

We developed the expert systein with Hitachi


workstation 2050/32 by using a software
programming tool "ESIKERXELIW". I t has a knowledge
expression of a hybrid type consisting of frames,
rules ("if-,then-" type) and C-language program.
The system structure and job selection of this
system on CRT display is s h o w i n Fig. 10.
Each knorvledge base item is divided into modules,
and the system has a tree-l ike structure.
This structure makes i t easy to add and revise the
knwledge bases.
After al I inference items are executed, the
inspect ion reports, rvhich include all items, are
printed out and present the results of the
insulation diagnosis quickly.

4. RESULUTS
The
comprehensive preventive maintenance
planning system has been actually applied to
maintenance plans of thermal and nuclear power
plants. Vie obtained good results by its
application. For example, a reduction of the
forced outage rate of nuclear power plants
(Boi 1 ing Water Reactor type plants by Hitachi Ltd.
as main contractor) improved from 0.3 times
/plant-year to 0.16 t imes/plant-year. The failure
rate in a thermal power plants also improved from
1.4 times / 6 months to 1.0 time/ 6 months.
Failure 'of a boiler tube after inspection ,
diagnosis and repair, has not occurred.

The comprehensive preventive maintenance


planning system has been developed Tor the
systematic, timely preparation of
recommendations on items used for periodical
maintenance and plant life extension of power
plants.
2. lhe diagnostic systems for remaining life
have been developed for turbine rotor,
generator coi I , rotor bar, pipe thickness,
and so on.
3.
The detection techniques for material
failure and remaining 1 ife assessment also
have been developed. That is:
(a) Combined assessment method using the
resu 1ts of electric resistance
measurement, rep1 ica and hardness.
(b) Estimation method of insulation
deterioration with the N-Y map and the
map method.
(c) Detection method of corrosion pitting.
1.

II

Fig. 10 System Structure and Job Selection of


Expert System

3.2.4

clpplicat ion f o r Generators

The remaining 1 ife estimation system of the


generator stator coils has been applied to 60
generators over a 10 year period. As a result of
diagnoses for coils which have already reached to
life expectancy or close to life, 21 coils of
polyester resin and 4 coils of epoxy resin in
generators have been rep laced. Generators which
have adequate remaining life rvill be diagnosed in
a period of four years.

152

The planning system can make a large


contribution to a preventive maintenance program
of power plants, especially in decreasing plant
forced outage rates with this system and its
continuous use.

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

REFEREMES

K. Onodera, Maintainabi 1 ity Design Techniques,


1989; Union of Japanese Scientists and Engineers
(JUSE .
2. S. Mori et al. Diagnostic Technique for
Remaining Life of Operating Power Plants, Hitachi
Review, Vol. 69-10, 1987 Oct. pp 83-88.
3. T.Tan et ai, Preventive Maintenance Technology
for Turbine Components, Hitachi Review,Vol.40,
1991 April, b . 2 . pp103-110.
4. S. h i , LTaniguchi, Diagnostic Technique
for Remaining Life of Large Rotation Coil, The
Proc. of 17th JUSE Reliabilty and Maintainability
Symp., 1987 June, pp 295-300.
5. K. Onodera, Expert System and Application for
Re1 iab i I i ty and Maintainab i 1 i ty, Re 1 iab i 1 i ty,
V01.9-3, 1987 NOV. pp 44-51.
1.

BIOGRAPHIES
Katsushige Onodera, PhD.
Hitachi Works, Hitachi. Ltd.
Operating Plant Services Center
1-1, Saiwai-cho 3-chome. Hitachi-shi, Ibaralii-Ken, 317 JAPAN
Katsushige Onodera is a mannger in the operating plant
services center. Since joining Hitachi horks in 1961, he has
been involved in re1 iabi I ity analysis and qual i ty assurance
activity in a nuclear power plant.
His PhD in engineering is from Tokyo University in 1990:
his B.S. in Electrical Engineering is from Yamagata
University in 1961.
He is a member of International Electrotechnical
Commission(1EC)- Dependability (IC-56),and a director of
Reliability Engineering Association of Japan(REAJ).

Yoomei Katoo, PhD


Hitachi Works. Hitachi,Ltd.
1-1, Saiwai-cho 3-chome.
Hitachi-shi, Ibaraki-ken, 317 JAPAN
Poomei Katoo is a chief engineer, Corporate technology in
Hitachi works. Since joining Hitachi,Ltd. in 1956, he has
been engaged in instrumentation and control technology
section of nuclear povier and chemical plant in Hitachi
Research Lab.
His PhD in engineering is from Osaka University in 1984:
his B.S. in electrical engineering is from Kanncawa
University in 1956.

1993 PROCEEDINGS Annual RELIABILITY AND MAINTAINABILITY Symposium

153

You might also like