Professional Documents
Culture Documents
A dissertation submitted
of
MASTER OF TECHNOLOGY
in
by
CHANDRAMOULI KASINA
Date : Date :
ACKNOWLEDGEMENTS
It gives me great pleasure to be able to thank everyone who has helped me see
this work completed. I am indebted to my thesis guides and supervisors, Dr Nesar
Ahmed & Dr Lajpat Rai for their valuable support and guidance throughout the
course of this work.
I sincerely thank Shri K.Prakasa Rao, GM, RSTPS, for allowing me to attend
the course work. I thank Shri S.K.Sekhar, Plant Manager, RRIPL, Kakinada for
extending valuable support during my visit to SPGL, Kakinada.
(Chandramouli Kasina)
CONTENTS
Certificate i
Acknowledgement iii
Contents v
Abstract vii
List of symbols ix
List of figures xiii
Chapter 1 Introduction
Chapter 2 Literature Survey
Chapter 3 Mathematical models & software development
3.1 Thermal & centrifugal stresses
3.2 Rotor temperature distribution
3.3 Heat transfer coefficient of rotor surface
3.4 After-first-stage steam condition model
3.4.1 After-first-stage steam temperature
3.4.2 After-first-stage steam pressure
3.5 Prediction of steam parameters & load
3.6 The software
Chapter 4 Results & Discussion
4.1 Stability of temperature distribution calculation
4.2 Effect of variable material properties on thermal stress
4.3 Present & predicted stresses
4.4 Rotor life consumption computation
4.5 Recommendations & future work
Chapter 5 References
Chapter 6 Bibliography
ABSTRACT
Fossil fuel powered power plants are conventionally being used to meet the
base load requirement. While the individual unit size is increasing on the one hand,
the difference between day time and night time demands is also increasing on the
other hand, forcing units into variable load operation. Non steady state operations like
start-up, shut-down and large load changes induce large thermal stresses in thick
walled components like steam turbine rotors, casing and valves. A turbine start-up
has to simultaneously satisfy two contradictive requirements. One, that of minimizing
the time taken for start-up due to reasons of economy of generation and second, that
of limiting thermal stresses within allowable maximum limits to ensure full service
life of the turbine components.
The levels of thermal stress reached in the critical regions of high pressure and
intermediate pressure turbine rotors is an important factor determining the life span of
the turbine. The present work implements a thermal stress monitoring system for a
typical steam turbine rotor with bore. This can easily be extended to solid rotors The
system computes both thermal and centrifugal stresses, based on the condition of the
steam at the inlet of the turbine. Rotor temperature and stress distribution are
determined using an explicit finite difference model of heat transfer. The variation of
material properties of the rotor and thk convective heat transfer coefficient is also
taken into account. The system extrapolates the current temperature trend of the steam
and computes predicted stresses for a look ahead period of 20 minutes. The system is
also capable of computing the life consumption of the rotor for any stress cycle
crossing a preset threshold. The program with suitable changes can also be used for
stress analysis of super-heater and reheater outlet headers of HRSGs in a combined
cycle plant.
vii
LIST OF SYMBOLS
MY WIFE
SATYA
DAUGHTER
POOJA
.
CHAPTER - 1
INTRODUCTION
In recent years, the need to meet efficiently the ever-increasing demand for
electric power has markedly increased the capacity of each steam turbine plant This
has made it necessary to operate a power plant in a manner to cope successfully with
variations in the demand for electric power, particularly to deal with the difference in
the demand for electric power between daytime and nighttime Thus, a power plant
hitherto operated at base load or with a constant turbine power is nowadays required
to frequently iffect startup, shutdown and load changes. For example, according to a
mode of operation called "Daily Start-Stop", boilers are set on at 5 A.M. every
morning and the start-up of the turbines is completed at 8 A.M The turbine is then
operated through out the day time until 10 P.M.This cyclic starting-up and stopping
of the plant are repeated every day. In another mode called "Weekly Start-Stop", the
plant is started up, for example, at 8 A.M. on Monday and works through the week
until it is stopped at 10 P.M. on Saturday, which cyclic operation is repeated every
week
Supposing that a plant in which the start-up must be completed at 8 A.M fails
to start at that time, disturbance (in fkquency due t o shortage of power) may result in
the system concerned. This disturbance can be overcome only through increasing the
outputs of other plant or plants. Thus, it is strictly required that the plants must be at
latest started by the time expected, e.g. at 8 A.M. which in turn necessitates minimum
length of time for starting-up the turbine. In addition, when a thermal power plant is
tripped due to abnormal conditions in the plant or power system, the plant must be
restarted as soon as the source of the trouble in the plant or system is removed.
The portion of the rotor facing the labyrinth packing after the first stage of
both HP and IP sectihs of turbine (the part of the rotor which comes in contact with
steam of maximum velocity and temperature, as shown in figure 1.1), is subject to
maximum thermal stress when turbine is started up or subjected to a load variation.
The larger the rate of change of speed or load becomes, the larger the thermal stress.
Proper control of thermal stresses at these points is of utmost importance from the
point of view of life of the rotor. In practice, for safe operation of the turbine, a quick
startup and an abrupt load change are strictly forbidden.
Figure 1.1 Cross-sectional view of a typical single casing HP-IP sections of turbine
2
The long-term reliability and availability of turbines requires close attention
to, and recognition of, the impact of thermal cycles on the "life" of steam turbine
components. Excessive thermal transients imposed on turbine rotating and stationary
parts during startups, major load changes, and shutdowns can greatly reduce the life
of these components. Because the effect of each thermal cycle is usually in itself
small, the effects may not be readily apparent to plant operators during the event and
may not be recognized and properly considered in making day-to-day operating
decisions. Damages like packing rubbing, shell distortion or even cracking of turbine
parts, is possible in the long run. Long-term cumulative effects of excessive thermal
stress cycles can result in extended and expensive forced outages requiring major
repairs or component replacement. In many instances, a turbine may appear to start
without difficulty, in spite of improper startup. During the initial periods of service,
no definite trouble may be evident.
Thermal stress monitor calculates the current stress and predicted stress in the
steam turbine rotor and advises the operator of the optimum rates of rolling and
loading. It also calculates the consumption of rotor life. A typical startup diagram is
given below in figure 1.2.
Time
b m p i i m t Full lord,-{
~
The thermal stresses generated in the turbine are closely related to the
acceleration or to the rate of change in load and generally tend to become larger as the
acceleration rate becomes larger. During start-up of the turbine, rolling-up and
warming are effected alternatively. Thermal stresses are related also to the duration of
warming. More specifically, thermal stresses are increased during acceleration and
then are gradually decreased during the subsequent
..~~warming. The larger the thermal
stresses, the longer is the duration of warming.
Most high capacity modem steam turbines have a double shell HP section
making the measurements of even casing temperatures of the inner shell difficult and
highly maintenance prone. The non-availability of measurements leads to non-
availability of thermal stress control systems.
The application areas of thermal stress calculations are in three broad inter-
linked categories.
Hanzalek & Ipsen's [13] path breaking article discusses the limiting conditions
occurring in high temperature components of HP and reheat sections of turbine, pain
stop and control valves. The concept of start-up charts for the operation of turbines for
rolling and initial loading is discussed. A sample start-up chart is shown in figure 2.1.
This is offline scheduler from the designer's point of view. The life consumption of
turbine rotors is also discussed. These offline scheduling charts are followed even to
the present day. The performance of all improved start-up schedulers is compared
with this as the base reference. References [20] and [21] represent the type of
information made available by the turbine manufacturers to the utility operators. [20]
represents a system where an off-line scheduler is similar to the one presented in [13].
Berry and Johnson 1171 discussed cyclic then@ stresses resulting eom
transient temperature changes and the consequent fatigue damage. They also
introduced the concepts of cyclic capacity charts. The calculations involved in the
grooved cylinder thermal stress is described in the annexure to the paper.
Fig 2.1 Turbine startup-chart
Adinarayana and Sastri [l] have reported the results of experiments conducted
on model steam turbine at BHEL, for the estimation of convective heat transfer
coefficients in industrial steam turbines. The results can be used in case of utility
turbines as well. They have shown that the heat transfer coefficient does vary a lot
during the rolling of the turbine and have stated that the semi-empirical relation of
Guzley[l8] needs improvement.
Matsumura eta1 [9] disclosed the aspects of a turbine start-up system with
influence of thermal stresses applied to a daily start-up and shut-down (DSS) unit.
The work closely follows the philosophy annunciated in [S].
. ..~
System designer's perspective of HITASS is presented by Honda[ll]. It
presents a block diagram (with no technical details) about the modeling of the after-
fust-stage steam parameters. This is more like the wnunercial pamphlet of HITASS
system.
The thermal stress of the rotor, i.e. the rotor surface thermal stress crn and
rotor bore thermal stress DBT are given by the following equations, on the basis of the
temperature distribution of the rotor aoss section.[l6]
The net stress in the rotor is finally calculated taking also the centrifugal stress
into account. Since the centrifugal stress is proportional to the square of the turbine
speed N, the centrifugal stress QC acting on the rotor bore at a turbine speed N is
given
where N is the speed of the turbine and OBCR is the centrifugal stress at the bore at the
rated speed NR. Consequently, the bore stress CB is given as follows.
Of the stress acting on the rotor surface, depending on the form of the rotor
surface, the thermal stress acts in the axial direction of the rotor, i.e. at a right angle to
the centrifugal stress which acts in the circumferential direction. Therefore, the
evaluation of the stress at the rotor surface necessitates only the thermal stress which
is concerned with the consumption of the turbine life. Thus, the stress os in the rotor
surface is given by
as = OST (3.05)
Thermal stresses can be either tensile or compressive depending on the temperature of
the surface(bore) being more or less as compared to the mean temperature of the
rotor. With present adopted convention, compressive stresses are negative in
magnitude and tensile stresses are positive in magnitude.
where OX 6t DBL are the maximum allowed surface and bore stresses and as% a~%
are the surface and bore stresses converted to percentage.
..
The governing equation for the heat conduction in a thick walled cylinder in
cylindrical co-ordinates is
where pis the diffisivity of the material. Two models suitable for numerical solution
of the above equation with suitable boundary conditions for turbine rotor are given
below.
The movement of heat in the rotor takes place predominantly in the radial
direction. The axial conduction is neglected. A cross-sectional view of the rotor is
shown in figure 3.1. The rotor is divided into m (1, 2, 3 . . . m) imaginary annular
sections . The temperature distribution is calculated by way of heat balances over the
annular sections.
QJs represents the heat delivered from the steam to the rotor surface in the
period 71. Similarly, QJI represents the amount of heat delivered from the rotor
surface to the core of the outermost G=1) annular section and QjJ.+lrepresents the
amount of heat delivered from the ja annular section to the j+1" annular section.
Since the rotor bore is kept in adiabatic condition, the heat amount Qm,,+r is always 0.
where AM is the thermal conductivity ofthe rotor material, K(t) is the rotor surface
heat transfer coefficient at instant t, Ts is the temperature of the rotor surface, rl is the
outer radius of the$ annular section, ro = rs is the rotor surface radius, r,+l= r~is the
radius rotor bore, Ar is the thickness of the annular sections, and represents the
1 Fig 3.1 Cross-sectional view of rotor
temperature ofJ* annular section. The heat transfer coefficient K(t), (calculated using
the model described in section 3.3), is used for calculating the heat amount QZs.
The amount of heat AQ,(l) accumulated in jth annular section is given as the
difference between the heat input Q,.iJ and the heat output Q,+i to and from the same
sectionJ
AQ# =Q,-/jCt)- Q,j+lC) (3 14)
The temperature T, of thejrhannular section is given by
where, 6 is the volume o f j h annular section per unit length, is the density of the
rotor material and CM is the specific heat of the rotor material. The rotor bore
temperature TB(t)is given by simulating the temperature distribution by a second
degree equation as follows.
dl;
-
-
- P <T,+~q..,) +
- P (?+I- 2Tj
---- + Tj-l 1
dt 2RjAr (w2
Figure 3.3 shows the method of calculating the convective heat transfer
coefficient K between the after-first-stage steam and the turbine rotor surface. The
modeling of the after-&-stage steam parameters is explained in section 3.4.
Block 4 calculates the flow rate FM (Kglsec) of the steam flowing through the
gap between the labyrinth packing and corresponding portion of the rotor using
Martin's equation.[9]
(G).
The constant k is characteristic
.. of a given turbine and is equal to bcsd.
PEl
Block 5 calculcttes the volumetric flow rate FV (m3 Isec) of steam flowing
through the gap between thellabyrinth packing and the rotor.
Block 6 calculates the velocity Urn (dsec) of the axial velocity of the steam
passing through the gap between the labyrinth packing and the rotor from the flow
rate Fv as
where A is the annular area (m2) between the labyrinth packing and the rotor. Block 7
calculates the rotor surface velocity URD(dsec) of the portion of the rotor at the
labyrinth packing.
where Rs is the surface radius of the rotor, and N is the speed of the rotor (RPM). The
combined leakage velocity U (mtsec) is then calculated as under by block 8.
R,O.B
are calculated by blocks 9 and 10 respectively. Finally, block 11 calculates
by 35.2
the heat transfer coefficient K ( K W I ~ ~ ~ofCthe
) heat transfer from the steam to the
rotor surface around the labyrinth packing behind first stage.
The pressure ratio 4 of the pressure after the second stage to the pressure after
the first stage is treated as a eonstant, because this ratio can be regarded as being
constant irrespective of the change of running condition, e.g. speed, acceleration and
load.[22]
I Rotor speed /
LZ-i1 temperature
* : - v - +
1
+ 2
v
I
Specific Coefficient of
weight kinetic
viscosit
1 1 4
4
Labyrinth packing
leakage flow
(Martin's equation)
I
t,
I5
Leakaage
volume flow
-
7 6
1
Rotor Axial leakage velocity
surface velocity
8
+ + ..
Combined leakage
velocity
9
+ v
Reynold's
Number
Nusselt
Fig 3.3 Computation of rotor heat transfer coefficient Heat Transfer Coefficient
(Gazley's equation)
Buzzi and Pasquantonio have suggested the following simpler empirical
relation for K.[24]
Figure 3.4 shows the method of obtaining the turbine steam parameters after
expansion through the first stage f?om the boiler outlet steam parameters. The method
uses main steam temperature %a, main steam pressure Pm, reheat steam temperature
Tm, speed of the turbine N, acceleration of the turbine N and load of the machine L
as input variables. The method is applicable for the entire range of turbine operation
i.e., rolling as well as loading. In case of after-first-stage steam temperature of
intermediatepressure turbine, it is assumed that there is no temperature drop across
the first stage of intermediate-pressureturbine. It is taken equal to the actual measured
reheat steam temperature. The IP section of the rotor being the most critical area for
stress related failures, this assumption also works as a safety measure. Further, load L
is assumed zero in no-load running condition (detected by the main breaker closing).
Once main breaker is closed, speed N is sssumed to be N, and acceleration N to be
zero. The model used for the determination of after-first-stage parameters is shown in
figure 3.4
The equivalent load L' equals zero during the rolling phase of the turbine.
------a- ~
~
Load -
N
N
T ~ S - Equivalent load under Function between load and
rated steam conditions after-first-stage steam temperature
~~.. ~
-
~ ~~ ~
~~~~ ..~ ~~~~~~~.
~-
Temperature drop
across control valve -. -.
.
Steam temperature
~
~~ .~ ~~
. ~~ ... ~~ ~.~~
. . ~ ~
~ ~
,8
7 - Mer-first-stage
-7 + steam pressure
D- ARer-first stage steam pressure ..~-. . ~~ ~
controls start with full-arc and switch to partial arc to reduce throttling losses through
the valves.
is then
calculated by block 2 from the characteristic of the turbine A typical characteristic is
shown in figure 3 5 The loads L l and L2 represent the lower and upper loads in case
This characteristic is also used for finding out the steam throttling ratio KI of
the turbine inlet control valve. The factor K l is also called coefficient of iris effect.
This is done by block 3. If the equivalent load L' is less than L2, then
. .~.
Block 4 determines the temperature drop ATo between the main steam
temperature TMSand the steam temperature in the turbine bowl. It is calculated &om
P.WS and lifs. This again is dependent on the configuration of the turbine. Typically,
ATo becomes larger as the pressure PMSbecomes greater, for a constant TMS.A typical
relation between ATo, PUS and Tm is shown in figure 3.6.[22]
Block 5 determines the temperature reducing factor K2 across the first stage of
the high pressure turbine. It is calculated as a function of turbine speed N.
Note that 0 5 K 2 S 1 and equals I when the turbine is operated at the rated speed and
during the loaded operation of the turbine.
24
Finally, block 6 finds out the after-first-stage steam temperature TH/of HP
turbine as
- q )+(Kt + K 2 -l)ATOR
THI=TMS-KIATo -K2(TMSI( (3.32)
The temperature Ti<, is determined as a value obtained by subtracting the temperature
drops of the steam on the way to the portion after-first stage, from the main steam
temperature TMS. K,AT,, represents the temperature drop across the turbine inlet
control valve, while K,(T,, - T, ) is the steam temperature drop across the first
stage. A T , represents the temperature differential between the main steam
temperature and the steam temperature in the turbine bowl under the rated steam
condition[9].
where KNLdenotes the after-fist stage steam pressure of the high-pressure turbine
corresponding to the no-load pressure drop at the rated turbine speed NR k denotes a
no-load pressure drop index number, and KAC denotes the pressure required for
obtaining a unit acceleration. Block 8 calculates the afier-first-stage pressure PHI
using
where P m is the after-first stage pressure at rated turbine conditions and L is the
load.
After-first stage steam pressure PI] of the intermediate-pressure turbine is
obtained (block 9) by multiplying the pressure PHIby the ratio of the behind-first
stage steam pressure PHIRof the high-pressure turbine at the rated load and the
corresponding PIIRof the intermediate-pressure turbine.
The same process is applied also to both the high-pressure and intermediate-pressure
turbine, with suitable substitution of the parameters/constants.
The formulation for thermal stress cited above can be used for computation of
both present and predicted stresses, depending on the parameters used. If the present
main(reheat) steam pressure and main(reheat) steam temperature are passed on to the
after-fust-stage steam condition model (discussed in 3.4), what we get is the present
temperature distribution of the rotor, which enables determination of present thermal
stresses (at both surface and bore). If, on the other hand, the expected rnain(reheat)
steam pressure and main(reheat) steam temperature (the steam condition expected at
the inlet of the turbine, if the present acceleration / loading is continued for a given
time) are passed on to the after-first-stage steam condition model, the predicted
stresses can be determined. For accurately .finding out predicted stresses, it is
necessary that the changing steam condition is accurately learned by the stress
calculation system.
and-).dP, The former three parameters are used in the speed control mode, while the
GP
latter three are used in the load control mode. The learning is done at a time interval
of r n q an integral number of stress computation interval using regression . The
following equations depict how the rate of change of a particular parameter is
calculated.
dTMS
-- _ Tm(t)-Tm(t-mrl) (3.37)
dN N ( t ) -N(t -m z , )
From the values of the rates of change of steam parameters, the predicted
values of steam parameters are obtained using equations (3.37), (3.38) and (3.39)
during the rolling phase and using (3.40), (3.41) and (3.42) during the loading phase.
d P .
P,(t+n~,)=P,(t)+~Ln~, (3.48)
dL
Here nz, is the prediction interval, L is the rate of change of load and Nis the
acceleration of the turbine.
3.6 THE SOFTWARE
The flow chart shown in figure 3.7 illustrates the method of estimation
of the rotor thermal stresses. The source code is listed in annexure B. At entry point
A, the condition of main steam (pressure and temperature at the inlet of control valve)
is available. Based on these parameters and the model of section 3.4, after-first-stage
parameters of steam are calculated by block 2. Having got the after-first-stage steam
condition, block 3 calculates the rotor heat transfer coefficient using model of section
3.3. The temperature distribution across the cross section of the rotor as shown in
figure 3.1 ( for assumed number of annular sections of 10) is calculated using the
formulation given in section 3.2 by block 4. Based on the temperature distribution
(rotor surface Ts, bore temperatures TB, and. .the temperature of the each of the
intermediate annular sections I;) the rotor mean temperature TMis calculated using
equation 3.16
The surface stress 0 .and bore stress UBS are then calculated by block 5 as
described in section 3.1 The program facilitates use of rotor material properties
either at a fixed user-defined temperature or as a function of temperature. Annexure-
A lists the properties of material 30CrMoV-as a function of temperature. The exit
point C gives the thermal stresses of th%r%oc given the geometry of the rotor,
',>V *
material properties of the rotor, CQIJ&&I&$~&&=~ inlet of the turbine, and the
temperature profile of the rotor at the-previous corhputation inieml. The same routine
is also used for calculating the predicted t h e m 1 stresses. In this case the main steam
,;
P.
parameters are made available at entry point B; ~nsteadof A.
Pre-estimate
steam condtiotion
at turbine inlet
2
Calculate
after-first-stage steam
condition
Calculate
rotor surface
heat transfer coefficient
..
7
4
Calculate
rotor temperature distributioon
Fig 3.7
+ swfaq &bore stresses
It is assumed that a linear temperature gradient is established along the radius of the
turbine rotor. This way equation 4.03 can estimate the temperatures T, of respective
sections of the rotor with a high level of precision, when the turbine is started after a
sufficiently long shutdown, because the difference between the temperatures TCOand
TCIis sufficiently small in such a case.
% 3'
.-
However, when the turbine is restarted after a short shut-down or the stress
monitoring system is brought into service with the turbine in hotlwarm condition, the
temperature distribution:& %turbine rotor is not estimated exactly because the
.- ,
..,,,,.-z..
Get process Data
Initial temperature
distribution
Steam condition
Interruption
by operator?
-'-I
Determine the stress l i i t s
Flow chart of figure 3.8 gives the steps involved in the main software
routine. Block 1 gets the casing temperatures on start-up of the system. Block 2
estimates the initial temperature distribution as per equations 3.49, 3.50 and 3.51.
Thermal stress computation is done once every minute. At each crossing of the
minute, Block 3 gets the main steam parameters. Block 4 keeps track of the main
steam parameters. his facilitates prediction main steam parameters by block 1 of
stress computation routine. The prediction routine, which works on linear regression
principle, will not function for the first 20 minutes, when enough data is not available.
Based on the casing temperatures, the maximum limits of stress are corrected in block
5. If ITco-Td < AT, the limits set by the designer 1 operator are used. If not, a
correction is applied to the limits. If G-L is the limit set by the designer, the new limit
becomes DL +Am where A o is a time-dependant correction. In the present program
the value of A o is taken as 10% of DL at the time of system --up and is linearly
reduced to zero over a period of 3 0 minutes. This means that at the end of 3 0 minutes,
the designer set limit or.becomes operational.
.-.
Blocks 6 & 7 calculate the present and predicted stress at the surface and bore
of the rotor (fig 3.7). Block 8 signals if any of the calculated stresses cross the limits
set by the operator I designer. The program can be made to log the stress values for
historical storage purpose.
Life of a rotor is dependent on stress values that are cyclic. By cyclic stress
values, it is meant that rotor stresses may assume positive (tensile) or negative
(compressive) values depending upon whether the stresses are caused by heating or
cooling. The positive maximum stress and a negative maximum stress determine the
stress range for a cycle. Block 9 calculates the life consumed by the rotor as shown in
e
figure 3.9. The range of stress the rotor surface undergoes is calculated by taking the
C ,ram Str-
Pmgrarn
Monitor
.-.
Fig 3.10 Typical stress variation
34
CHAPTER - 4
time (MIN)
time (MIN)
The turbine rotor is made of special steels like 30CrMoV. Its properties
Young's modulus (9,Poisson's ratio (v), coefficient of linear expansion (a),specific
heat of rotor material (CM), heat transfer coefficient between steam and rotor (K),
thermal conductivity of the rotor material (AM)and density of the rotor material b)
are all temperature dependant. A simulation of the variation of
rotor mean temperature,
surface stress and
bore stress
is run by taking rotor material properties at fixed temperatures of I00 deg C, 300 deg
C and 500 deg C and also variable material properties. The variation of rotor mean
temperature is shown in figure 4.4. The fixed temperature material property curves
fall on either side of variable properties curves. To see the trend of variation, the
deviation of mean rotor temperature calculated with fixed temperature properties of
100, 300 and 500 deg C with reference to mean rotor temperature calculated with
variable material properties is plotted in figure 4.5. Similarly, the plots of variation of
surface and bore stresses, and their deviation with respect to stresses calculated with
variable material properties are plotted in figures 4.6,4.7,4.8 and 4.9.
It is observed that the computation done with fixed material properties at 300
deg C comes closest to the computation with variable material properties. This is
especially true in case of bore stress, which is the most important stress to be
monitored. It can be concluded that in a stress monitoring system implementing fixed
material properties, the errors in computation can be minimized by using material
properties at around 300 deg C (approximately midway between 100 deg C and 500
deg C ,the usual lower and upper limits of after-first-stage steam temperature).
Variation of rotor mean temperature
Effect of rotor rnateral properties
i I-
afs-temp -100 deg coeffs - J X ) d e g coeffs -5COdeg coeffs -var temp coeffs )
I
.,, 1
time (MIN)
-100 deg coeffs -333 deg cwffs -500deg Meffs --va~iabletemp weffs /
I
. - -
time (MIN)
-120
-1 40
-160
time (MIN)
0 C
0 20 40 60 80 1CO 120 140
time (MIN)
100
80
60
s 4
Me 20
6 o
E
$ -20
40
-60
80
time (MIN)
I
Fig 4.9Variation of rotor bore stress deviation
4.3 PRESENT & PREDICTED STRESSES
The variation of present and predicted stresses at both surface & bore for the variation
of after-first-stage steam temperature as shown in figure 4.10 are plotted in figure
4.11 and 4.12 respectively.
Present IPredicted stresses
(Prediction intewal - 20 Min)
Of late, "Triplex Modular Redundant" (TMR) systems are being used for
steam turbine control, where three processors are used for control purpose with a
voting logic built-in between the processors. In systems like this, the execution time
of the control algorithm is of critical importance and should be as small as possible.
Calculating rotor bore and surface temperatures based on the time response of the
rotor and fixed rotor material properties is a reasonable compromise so as not to
overload the cycle time. The present software can be used for finding out the time
response of the rotor mean temperature with respect to the rotor surface temperature.
Plots like the ones shown in figure 4.4 can be used for finding out the rotor response
characteristics.
The software is capable of finding the Life consumption based on the range of
stress the rotor is subjected to by monitoring the surface stress variation. Figure 4.13
shows the variation of surface stress. A dead band of +50% is assumed. The portions
of surface stress crossing the dead band is monitored for maximum and minimum
stresses. The range of stress change is computed by taking the difference between
maximum positive and negative surface stresses. This is converted to percentage life
consumed from the manufacturer supplied characteristic.
i Life consumption calculation
time (MIN)
REFERENCES
Seong Woo Kwak and Byung Kook Kim, "A study on automatic control of
steam turbines of fossil power plant with thermal stress constraints in turbine
rotors", IECON proceedings (Industrial Electronics Conf) Vol 2, 1995, Pp
1525-1530
Young WC, "Roark's formulas for stress and strain'', McGrawHill, 1989
Manson SS, "Thermal stress and low cycle fatigue", McGrawHill, 1966
Parsons Power Generation Ltd, "Thermal Stress Monitof' for Godavari Gas
Power Project.
Personal communication
Nuyak & Dua, 'Thermal analysis of steam turbine casings under steady state
and transients"
CHAPTER - 6
BIBLIOGRAPHY
Kamiya, A,; Kakei, A.; Kawai, K.; Kobqmhi, S, "Advanced power plant start-
up automation based on the integration of soft computing and hard computing
techniques", Systems, Man, and Cybernetics, 1999. IEEE SMC '99
Conference Proceedings. 1999 IEEE International Conference on , Volume: 1
, 1999, Page(s): 380 -385 vol. 1
Kamiya, A.; Kobayashi, S.; Kawai, K., 'Xeward strategies for adaptive start-
up scheduling of power plant", Systems, Man, and Cybernetics, 1997.
Computational Cybernetics and Simulation., 1997 IEEE International
Conference on, Volume: 4 , 1997, Page(s): 417 -3424 vo1.4
Smith CA et al, 'Tuzzy and model based supervisory controller for combined
cycle power plant start up using thermal stress constraints", 3'* European
control conference ECC '95 Rome, Pp 2444-2449
Bednarski S and Shen CN, "Analysis and algorithms for min-max problem
with thermal stress application", 14&Joint Automatic Control Conference, Vol
20, No 22,1973, Pp 765-775
IEEE Commitfee report, 'mynamic Models for steam and hydro turbines in
power system studies", IEEE trans on Power Apparatus & Systems, Vol 92,
NO 6, Nov/Dec 1973, Pp 1904-1915
ANNEXURE - A
PLANT DATA
stress calculation
maxSurfStress Maximum safe limit for surface stress MPa
maxBoreStress I Maximum safe limit for bore k e s s IM P ~
dbstress Safe band of stress beyond which life consumption is Mpa
to be calculated
ratedspeed Rated speed of the turbine RPM
//
void firstStageParams ( double mspress, double mstemp,
int speed, double accel,
double load, int genBrkr,
double *press, double *temp);
void InitialDistrib (double cas100, double cas50);
void stress(doub1e *surfaceT, double *boreT, double *meanT,
double *stressl, double *stress2);
void presentStress(doub1e *surfaceT, double *boreT, double *meanT,
double *presentSS, double *presentBS);
void predictedStress(doub1e *predictedSS, double *predictedBS);
ofstream outfile("allstr.txt");
void main ( )
I
double presentSS=O.O,
presentBS=O . 0,
predictedSS=O.O,
predictedBS=O. 0;
r(O] = raddurface;
r[l] = radsurface - deltaR/2;
for (i=O;ii=20;i++)
speedprev = speed;
loadPrev =load;
for (i=O;i<ZO;i++)
histP[201 = mspress;
histT[201 = mstemp;
histN[Z01 = speed;
histL[ZOI = load;
else
i'
tempslope = linReg( histL, histT, 10);
pressslope = linReg( histl, histP, 10);
multiplier = 1Dot;
1
deltasigmasurface = max(sigmaCorrectionMag*(l-
noMin/sigmaCorrectionTime),O);
deltaSigmaBore = max(sigmaCorrectionMag*(l-
noMin/sigmaCorrectionTime),O);
if (noMin<=towP)
{ ..
tempPred = mstemp;
presspred = mspress;
)
else
for (i=l;i<=noseg;i++)
1
tPresent[i] = tPresent[O] - (tPresent[O] - tPresent[lll) *
(2*i-1) / (Z*no.Seg);
tPredicted[i] = tPredicted[O] - (tPredicted[Ol -
tPredicted[lll) * (2*i-1) / (2*noSeg);
1
for (i=O;i<=noSeq+l;i++)
1
t[i] = tPresentIi1;
I
afs-temp = mstemp; ..
afs-press = mspress;
for (i=O;i<=noSeg+l;i++)
*surfaceT = surfaceTemp;
*boreT = boreTeinp;
*meanT = meanTemp,
*presentSS = presentSS1;
*presentBS.= presentBS1;
for (i=O;i<=noSeg+l;i++)
for (i=l;i<=noSeg-l;i++)
(
/ / amount of heaat delivered from segment i to segment i+l
q[il = 2* pi * r[i+l] * lamda(t[i]) (t[i] - t[i+l] ) calPer
I
dqLnoSeg1 = q[noSeg-11;
t [noSegl = t [noSegl + dq[noSegl / cp (t[noSegl ) / rho (t[noSegl /
(2* pi * r[noSegl * deltaR);
tsurface = t [O];
tBore = t [noSeg+ll;
stressconstant = E(tMean) alpha(tMean)/(l-nu(tMean)J;
sigmas = stressconstant * (tMean -
tSurface)*100/ (maxSurfStress+deltaSigmaSurface);
sigmaB = stressconstant * (tMean -
t~ore)*100/(maxBoreStress+deltaSigmaBore);
+surfaceT = tSurface;
*boreT = tBore;
*meanT = tMean;
*stress1 = sigmas;
*stress2 = sigmaB;
if (fixCoeffFlag==l) ..
return (fix);
else
return (var);
!
if (fixCoeffFlag==l)
return (fix);
else
return (var);
double nu (double temp)
/ / find poisson's ratio as a function of temperature
I
double fix,
var,
XI] = {0,100,200,400,600),
y[] = {0.3,0.3,0.3,0.3,0.3);
/ / dimension
less
//nu = 0.3;
fix = lagInter (x,y,5,fixCoeffTemp);
var = laqInter(x,y,5,temp);
if (fixCoeffFlaq==l)
return (fix);
else
return (var);
1
//k = 24.01465;
fix = lagInter(x,y,3,fixCoeffTemp);
var = lagInter (x,y,3,temp);
if (fixCoeffFlag==l)
return (fix);
else
return (var);
)
//rho = 7850.0;
fix = lagInter(X,y,2,fixCoeffTemp);
var = lagInter (x,y , 2 , temp);
if (fixCoeffFlag==l)
return (fix);
else
return(var);
1
//cp = 485.518;
fix = lagInter(x,y,4,fixCoeffTemp);
var = lagInter(x,y,4,temp);
if (fixCoeffFlag==l)
return (fix);
else
return(var);
1
double kl = -7.691234564,
k2 = -26.08023696,
k3 = -168.1706546,
k4 = 64.23285504,
k5 = -118.9646225,
k6 = 4.167117320,
k7 = 20.97506760,
k8 = 1.0e9,
k9 = 6.0;
..
double beta, theta, / / normalised pressure & temp
x, nul, denl, den2; / / intermediate values
double LO
L1 -= 15.74373327,
-34.17061978,
I1 = 4.260321148,
a1 = 8.438375405e-1,
a2 = 5.362162162e-4,
a3 = 1.72,
a4 = 7.342278489e-2,
a5 = 4.97585887e-2,
a6 = 6.5371543e-1,
a7 = 1.15e-6,
a8 = 1.5108e-5,
a9 = 1.4188e-1,
a10 = 7.002753165,
all = 2.995284926e-4,
a12 = 2.04e-1.
All = 7.982692717,
A12 = -2.616571843e-2,
A13 = 1.52241179e-3,
A14 = 2.284279054e-2,
A15 = 2.421647003e2,
A16 = 1.269716088e-10,
A17 = 2.074838328e-7,
A18 = 2.17402035e-8,
A19 = 1.105710498e-9,
A20 = 1.293441934e1,
A21 = 1.308119072e-5,
A22 = 6.047626338e-14,
else
T = temp+TZA;
tRat = T/TCA;
lamda0 = pow(tRat,0.5) * (a0 + tRat* (a1 + tRat* (a2 + a3
tRat)));
rho = l/spvol(press,temp);
rhoRat = rho/rhoStar;
lamdaBar = bO + bl rhoRat + b2 * exp(Bl*pow(rhoRat+B2,2) ) ;
dtStar = fabs (tRat - 1.0) + C4;
Q = 2.0 + C5 * pow(dtStar,-0.6);
R = Q + 1.0;
if (T < TCA)
S = C6 * pow(dtStar,-0.6);
else
S = pow(dtStar,-1.0);
term1 = (dl* pow(tRat,-10) + d2) * pow(rhoRat,l.B) * exp(Cl*(l-
pow(rhoRat,2.8) 1 ) ;
term2 = d3* S * pow(rhoRat,Q) * exp( (l-~ow(rho~at,~)*~/R));
term3 = d4 * exp(CZ*pow(tRat,l.5) + C3 * pow(rhoRat,-5));
deltalamda = term1 + term2 + tenn3;
return (lamdaO + lamdaBar + .deltaLamda);
I
double muSteam(doub1e press, double temp)
(
/ / unit press MPa
// temp degC
// mu Pa * le-6
rhostar = 317.7,
rho,rhoRat, T, tRat, terml=O, term2=0,
mu0, mu;
rho = l/spvol(press,temp);
rhoRat = rho/rhoStar;
T = temp + TZA;
tRat = T / TCA;
1
mu0 = pow(tRat, 0.5) / term2;
mu = muO*exp (rfioRat*termll;
return (mu);
return (muSteam(press,temp)*spvol(press,temp));
)
I
double nDot, / / acceleration of turbine
n, / / speed of turbine
L, / / load of turbine
Ldash, / / load corrected to rated
parameters ..
tldash, / / after-first-stage temp at
Ldash
xLdash[l = {O, 20,70, 1001, / / Ldash characteristic
ytldash[l = 1270, 450,425, 4951.
L2, / / upper limit of combined governing
kl, / / steam throttling ratio
k2, / / temp reducing (cooling)
factor
deltaT0, / / ms temp and turbine bowl
steam temp dt
deltaTRO = 40.0, / / rated ms temp and turbine
bowl steam temp dt
pMIN = 7.0, / / min pressure in deltaTO
characteristic
pHlp = 13.2, / / rated afs pressure
kAC = .001, / / pressure required for unit
acceleration
kNL = .05, / / afs pressure at rated speed
& no load
k = 1.2, / / no load pressure drop index
number
xtMS[I = (295, 400, 535),// deltaTO characteristic
ydeltaT0[1 = !60,50,40),
PMS , / / ms pressure
tMS. / / ms temperature
pMSR = 15.0, / / rated ms pressure
tMSR = 535.0, / / rated ms temperature
~10, / / a55 pressure at no load
pH1, / / afs pressure
tH1; / / afs temperature
pMS = mspress;
tMS = mstemp;
no = ratedspeed;
if (qenBrkr == 1)
1
nDot = 0.0;
n = ratedspeed;
L = load;
1
else
{
nDot = accel;
n = speed;
L = 0.0;
1
else
1
kl = (L2 - Ldash)/L2;
..
1
Area= (pow(radsurface+clearance,2)-pow(radSurface,2~~/~le~~;
specificweight = l/spvol(press,temp);
kineticviscosity = nusteam(press,temp);
heatconductivity = lamdaSteam(press,temp);
massFlowRate = kO * sqrt(press * specificweight) * sqrt((1-
pow(ph2ByPhl,2))/(Z-log(ph2ByPhl)l);
volFlowRate = mass~low~ate/specificWeight;
axialVelocitySteam = vol~low~ate/Area;
surfaceVelocityRotor = 2 * pi * radsurface * float(speed/60);
effectivevelocity =
sqrt(pow(axial~elocity~team,2)+(pow(surface~elocityRotor/2,2)));
Re = 2.0 * effectivevelocity * clearance / kineticviscosity;
Nu = pow(Re, 0.8)/35.2;
coeff = heatconductivity* Nu /Z.O/clearance;
if (fixCoeffFlag==l) return (100001;
else return (coeff);
1
..
void inputparams( double *rnspress, double *=temp, int *speed, double
*load, int *gtnBrkrl
I
// include code for simulation of various parameters
/ / define the break points for each paraemeter as a function of
time
/ / and used Lagrandian intrpolation routine for generation of
/ / various parameters