You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267620219

Monitoring of HRSG Performance in Large Size


Combined Cycle Power Plants

Conference Paper · January 2008


DOI: 10.1115/GT2008-50648

CITATIONS READS

2 336

3 authors, including:

Traverso Alberto Aristide Fausto Massardo


Università degli Studi di Genova Università degli Studi di Genova
91 PUBLICATIONS 984 CITATIONS 167 PUBLICATIONS 2,559 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

resilient polygeneration RES smart grids optimisation View project

Third generation pressurised Hybrid Systems based on SOFC technology View project

All content following this page was uploaded by Aristide Fausto Massardo on 25 July 2015.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Proceedings of ASME Turbo Expo 2008: Power for Land, Sea and Air
GT2008
June 9-13, 2008, Berlin, Germany

GT2008-50648

MONITORING OF HRSG PERFORMANCE


IN LARGE SIZE COMBINED CYCLE POWER PLANTS
Silvio Cafaro Alberto Traverso Aristide F. Massardo
silvio.cafaro@unige.it alberto.traverso@unige.it massardo@unige.it

University of Genoa
Thermochemical Power Group
via Montallegro 1, 16145
Genova - Italy

ABSTRACT analysis, each performance parameter is coupled with an


accuracy factor. The accuracy of each performance parameter is
Monitoring of all components of large size combined estimated through the sensitivity analysis, which allows to
cycle power plants (gas turbine, HRSG, steam turbine, determine the best parameters to be monitored and to define the
auxiliaries) plays a determinant role in improving plant related tolerance due to measurement errors.
availability, profitability and maintenance scheduling. The methodology developed has been successfully applied to
This paper presents a research project carried out by TPG historical logged data (2 years) of an existing large size (400
(Thermochemical Power Group) of University of Genoa in MW) combined cycle, showing the capabilities in estimating
collaboration with Ansaldo Energia S.p.A. to improve existing the degradation of the HRSG throughout plant life.
monitoring and diagnostics procedures and to develop
innovative software tools for software-aided maintenance and INTRODUCTION
customer support: the first part of research is concerned with
the monitoring of a three pressure level HRSG (Heat Recovery Combined Cycle (CC) Power Plant performance analysis is
Steam Generator), which is presented in this paper. very important because of their large diffusion all over the
A procedure for estimating HRSG performance in large world and of their leading role in the electric power production.
size combined cycle power plants is presented. The work A complete performance analysis is essential both for the
consists of the development of an original Matlab code which power plant constructor and for the user.
calculates heat exchangers’ performance, at different power Most of the CC constructors offer a suitable plant monitoring
plant operating conditions. The Matlab code uses some system for each kind of plant they produce, with the objective
parameters (areas of heat exchangers, heat transfer coefficient, to control daily the performance, efficiency and operation of
heat loss, pressure drop) coming from a detailed on-design the whole power plant. As reported in [10], the monitoring
model necessary to set some parameters for the calculation. The activity consists in evaluating continuously the productivity
original Matlab code was developed with a twofold objective: capacity and the efficiency of the plant, using the stream of data
to calculate the actual gas path inside the HRSG starting from coming from plant instrumentation. The performance
the available measurements, thus obtaining the current evaluation is repeated at regular intervals with the objective of
effectiveness of all the heat exchangers in the HRSG; to calculating the degradation of the different elements of the
estimate the expected performance of each heat exchanger to be plant. The degradation is defined as a worsening performance
compared with the actual ones. and can be caused, depending on the plant component, by
Once the actual effectiveness and the expected various phenomena: from compressor fouling to turbine blade
effectiveness of the heat exchanger are defined, non- oxidation or corrosion, heat exchangers fouling, and so on [1]
dimensional performance parameters suitable for degradation [2] [3] [4].
assessment can be defined. Such parameters will be used to A detailed diagnostic analysis can perform important
monitor plant degradation, to support plant maintenance, and to improvements in power plant management, such as:
assist on-line troubleshooting. As a result of the sensitivity

1 Copyright © 2008 by ASME


• Instant information about the operation of the plant, Non Dimensional Performance Index based on
NDPIA
through the definition of efficiencies and degradation surface


indexes NTU Number of Thermal Unit
Troubleshooting: fast identification of the malfunctioning Nu Nusselt number
component and possibility to reduce the maintenance p Pressure [bar]
shutdowns of the plant or to avoid stops with targeted Pr Prandtl number


interventions Q heat [kJ]
R characteristic parameter of an heat exchanger

Performance improvement
Better management of maintenance (Predictive Re Reynolds number
sm element of the “sensitivity matrix”

Maintenance)
Optimization of plant operation T Temperature [K]
TOT Turbine Outlet Temperature [K]
In Figure 1 it is possible to observe a schematic representation TVA Turbine Vibration Analyzer
of the future HRSG monitoring system. U heat transmittance [kW/m2·K]

GREEK LETTERS

ΔxEM uncertainty of the measurement instrument


imposed (1%) uncertainty of the measurement
Δx SM
equipment
ε heat exchanger effectiveness
λ thermal conductivity [kW/m·K]
µ dynamic viscosity [kg / m·s]
ρ density [kg/m3]
φ constant

SUBSCRIPTS
Figure 1. Schematic representation of the HRSG Monitoring
system ACT Actual
Amb ambient
NOMENCLATURE EI Exhaust Inlet
EWI Exhaust Water Inlet
A area of heat exchange EO Exhaust Outlet
AEN Ansaldo Energia EOSW Exhaust Outlet Steam/Water
B constant EXP Expected
c flow velocity [m/s] in inlet
CC Combined Cycle max maximum
cp specific heat at constant pressure [kJ / kg·K] min minimum
D characteristic diameter [m] out outlet
em element of the “error matrix” REF Reference
the larger between the two thermal capacities which SWI Steam Water Inlet
Emax flow through the heat exchanger SWO Steam Water Outlet
the lower between the two thermal capacities which WI Water Inlet
Emin flow through the heat exchanger
f calculated output 1 REFERENCE COMBINED CYCLE
GT Gas Turbine
h specific enthalpy [kJ/kg] The type of power plant, which HRSG is the object of this
HR Relative Humidity [%] study, is a combined cycle, producing around 400 MW,
HRSG Heat Recovery Steam Generator composed by a gas turbine (Ansaldo Energia V94.3A2), a 3
IEE Index of Expected Error level HRSG (constituted by 13 heat exchangers), a steam
L characteristic length [m] turbine with 3 pressure expansion (high, intermediate, low
pressure), air or water condenser. Figure 2 shows a schematic
m& mass flow rate [kg/s]
plant layout, including calculation sections.
Non Dimensional Performance Index based on
NDPI
effectiveness

2 Copyright © 2008 by ASME


intermediate and high pressure lines have the same layout and
are composed of two economizers, one evaporator and two
superheaters.
As the cycle is reheated, there is also a reheater between the
first and the second high pressure superheaters. The mass flow
that crosses the reheater is the sum of the high and intermediate
pressure flows. A consistent part of the steam mass flow
leaving the reheater (10%-20%) is extracted and used for
cogeneration.
There are also 3 desuperheaters, used as temperature
controllers inside the HRSG, these are:
1. Extracted steam flow desuperheating: the set point
temperature of the extracted steam flow is 543 K
(extracted at this temperature because of process
requirements)
2. Upstream reheater desuperheating: the maximum
temperature at the steam turbine outlet must be 603 K
3. Desuperheating between the two high pressure
superheaters: set point temperature of the steam flow
Figure 2. Schematic representation of the plant (sections of temperature approaching the second superheater is 733 K
calculation) at nominal conditions. In different load conditions it
changes, but this desuperheating represents a regulating
1.1 GAS TURBINE parameter of the HRSG.

The V94.3A2 is an heavy duty, single shaft, axial gas 1.3 STEAM TURBINE
turbine, operating at 50 Hz; its fuel is usually natural gas and
the nominal power at ISO conditions is around 272.4 MW. The steam generated in the HRSG is sent to a steam
The V94.3A2 has the following characteristics: turbine, composed of two groups: high pressure steam turbine
and middle-low pressure steam turbine. It is coupled with a 50
- 15 stages axial-flow compressor Hz generator; the power at nominal conditions is around 135
- annular combustion chamber MW, while the steam admission pressure and temperature are
- 4 stages turbine respectively 115 bar and 811 K.

The most important gas turbine thermodynamic data are The following calculations were made using a real set of data
reported in Table 1: coming from an AEN monitored plant.

Turbine Model V94.3A2 UOM 2 OFF-DESIGN MODEL OF WHOLE COMBINED


Frequency 50 Hz CYCLE
Power output at terminals (ISO-Base load) 272.4 MW
Exhaust mass flow 682 kg/s The off-design model of the whole plant was developed
Exhaust temperature 852 K using the commercial code Gate Cycle®. The development of
an off-design model on Gate Cycle® was necessary for setting
Efficiency at terminals (ISO-Base load) 38.51 %
parameters useful in the calculation of the HRSG performance;
the code doesn’t need to run the Gate Cycle® model. First, the
Table 1: V94.3A2 thermodynamic data (Year: 2007)
GT model was set up, then the HRSG, steam turbine and
auxiliary models were added to have the full representation of
1.2 HRSG
CC performance. Gate Cycle® has a large library of several
commercial gas turbine; for this reason the module called
The HRSG produces superheated steam at 3 pressure
“GTDATA” was chosen for modelling the gas turbine (AEN
levels: 7 bar, 50 bar, 155 bar. The low pressure line is
V94.3A). In this way gas turbine correction curves can be
constituted by one economizer (which is also called “feedwater
customized, in order to have a more accurate model. This
heater”) that provides the first heating of the whole water flow
means that is possible to correct the calculated performance of
coming from the extraction pump, one evaporator (that
the gas turbine when a variation in the ambient conditions
produces the low pressure steam) and one superheater (that
occurs. The correction curves of the GT are shown in Table 2:
produces the superheated low pressure steam). The

3 Copyright © 2008 by ASME


TYPE OF CORRECTION CURVE • pressure drops
Exhaust Flow vs. Compressor Inlet Temperature
Exhaust Flow vs. Load (%) The calculation is developed applying continuity and energy
Exhaust Temperature vs. Compressor Inlet Temperature equations (at steady-state) at each section of the HRSG (see
Exhaust Temperature vs. Load (%) Figure 2).
Heat Rate vs. Compressor Inlet Temperature
Heat Rate vs. Load (%) 3.1 INPUTS
Power vs. Compressor Inlet Temperature
The inputs of the original Matlab code were taken directly
Table 2: Type of GT correction curve from the AEN monitoring system (TVA), selecting
instantaneous operating conditions of the whole power plant at
The validation of the gas turbine model was done using original steady state. The criteria used for the choice of the exact
data from AEN; the difference between the model and the moment for collecting the data were: stability of gas turbine
original data was less than 1.5% for each characteristic and steam turbine power, stability of steam flow extracted,
parameter (ex.: Power, Exhaust mass flow rate, Exhaust stability of steam pressures.
Temperature, Compressor Isentropic efficiency, Turbine 34 points were collected, about two points of measurement
Isentropic efficiency). Once the gas turbine model was every month, in a time interval of about 2 years; this time
validated, it was possible to construct the combined cycle interval was considered good for underlining a degradation in
model, putting together all the elements of it: HRSG (with the heat exchangers performance.
single heat exchangers), steam turbines, condenser, auxiliaries.
The validation of the on-design complete model was carried out 3.2 MINIMUM SET OF MEASUREMENTS
comparing the thermodynamic data from the “performance
test” with the data from the model, until acceptable fitting was Relating to Figure 2 the set of measurements used as input
achieved. The off-design model was validated comparing the to the code is reported in Table 3.
most important physical parameters of the cycle (e.g.: CC total
power, fuel mass flow, exhaust mass flow, TOT, high pressure IDENTIFIER TYPE OF MEASUREMENT
steam flow and temperature, process steam flow) with the T1f Exhaust Temperature
measurements. pf Exhaust Pressure
m1f Exhaust Flow
3 MATHEMATICAL APPROACH OF HRSG O2 O2 Molar Fraction in the Exhaust Flow
THERMODYNAMIC DIAGNOSTICS N2 N2 Molar Fraction in the Exhaust Flow
Ar Ar Molar Fraction in the Exhaust Flow
The “core” of the HRSG monitoring system is an original SO2 SO2 Molar Fraction in the Exhaust Flow
Matlab code, that calculates performance parameters of the heat CO2 CO2 Molar Fraction in the Exhaust Flow
exchangers and their related accuracy. H2O H2O Molar Fraction in the Exhaust Flow
All the calculations made in this code were done using real He He Molar Fraction in the Exhaust Flow
measurements from the plant.
m2 Water Flow
The objective of such a code is to calculate:
m3 Steam Flow
- all the exhaust temperatures (representing the gas path m10 Steam Flow
inside the HRSG) m16 Steam Flow
- the current effectiveness of all the heat exchangers in m19 Water Flow
the HRSG (called “actual” effectiveness) [8] m20 Water Flow
- the “reference” effectiveness of all the heat m21 Water Flow
exchangers in the HRSG [8] m22 Water Flow
- the NDPI of each heat exchanger mSPILL Steam Flow
- the IEE for each NDPI Tamb Ambient Temperature
T0 Steam Temperature
The parameters of the heat exchangers taken from the on- T2 Water Temperature
design model and implemented into the diagnostic model are: T8 Steam Temperature


T9 Steam Temperature


area of heat exchange T10 Steam Temperature


heat transfer coefficient T14 Water Temperature
heat losses T15 Steam Temperature

4 Copyright © 2008 by ASME


T17 Water Temperature An important assumption in the calculation tool is that the heat
T20 Water Temperature exchangers are considered counter-flow. All of them are locally
p0 Steam Pressure cross-flow but they can considered globally counter-flow.
p2 Water Pressure
p4 Steam Pressure Exhaust Inlet Exhaust Outlet
p8 Steam Pressure
Heat
p9 Steam Pressure
Steam/Water Outlet Exchanger Steam/Water Inlet
p10 Steam Pressure
p12 Steam Pressure
p14 Water Pressure Qloss
p16 Steam Pressure
p20 Water Pressure Figure 3. Schematic representation of a generic heat exchanger

Table 3: Type of measurement Figure 3 shows a generic view of a heat exchanger. From a
more detailed point of view it is necessary to differentiate
3.3 CONSERVATION EQUATIONS AT STEADY-STATE among economizers, superheaters and evaporators (and
deaerator), because the way of calculating the effectiveness
The equations used in the development of the code are the varies.
continuity and the energy equation at steady state. For economizers and superheaters it is:

(TSWO − TSWI )
The continuity equation imposes that the mass flow entering a

ε=
physical element is the same exiting the element itself.
(TEI − TSWI )

∑ m& = ∑ m& out


In mathematical terms it is possible to write: (3)

(1)
in
For evaporators:

(TEI − TEO )
ε=
(TWI − TEISW )
The energy equation imposes that the energy entering a
physical element is the same exiting itself, minus losses. (4)
For a heat exchanger two type of losses are considered:
convective loss between two following heat exchangers
(usually neglected) and convective loss between each heat Equations (3) and (4) follow from the counter-flow heat
exchanger and the ambient. Qloss is a linear function of the exchanger theory [11]: the upper temperature difference of ε
temperature difference between the average gas temperature must be the highest between the two fluids. In fact considering
and the ambient temperature. In mathematical terms it can be superheaters and economizers the higher difference between
written as: the input and output temperatures is the one of the water/steam

∑ m& in ⋅ hin − Qloss = ∑ m& out ⋅ hout


flux, while for evaporators the higher difference is between the
(2) inlet flue gas temperature and the outlet flue gas temperature.
The value of the “actual” effectiveness for each heat exchanger
will be calculated as illustrated before.
Setting these equations (in a matrix form) for each heat
3.5 NON DIMENSIONAL PERFORMANCE INDEXES
exchanger, the gas path of the exhausts in each section of the
HRSG can be calculated.
The Non Dimensional Performance Indexes represent the
most important result of the calculation: they give information
3.4 HEAT EXCHANGER EFFECTIVENESS
about the operating conditions of each heat exchanger of the
HRSG and in particular about their good operation.
Once all the temperature values of the gas along the
For example, a rapid change in the NDPI trend of a heat
HRSG are known, the “actual” effectiveness of each heat
exchanger would probably mean that there is a leakage in a
exchanger can be obtained.
pipe or that a pipe of the heat exchanger is broken.
The effectiveness of a generic heat exchanger is defined as the
The NDPI are defined through the comparison between the
ratio between the real heat flux exchanged and the theoretical
“actual” effectiveness (calculated as described before) and the
maximum heat flux exchanged in an ideal heat exchanger.
“expected” effectiveness, as mathematically described in (5):

5 Copyright © 2008 by ASME


ε ACT
NDPI = ⋅ 100
ε EXP
(5) Where the direct formula is (9):

1 − e − NTU *(1−R*)
ε REF =
1 − R * ⋅e − NTU *(1−R*)
NDPI unit of measure is percentage [%]; in this way NDPI = (9)
100% means that the heat exchanger is performing at its best,


and it is considered new and clean. Such indexes require the
definition of the “expected” performance of a heat exchanger. Calculation of the heat exchanger reference
The “expected” values are calculated using a few parameters transmittance UREF

NTU REF ⋅ Emin


derived from the Gate Cycle® model, but without the need of

U REF =
running it, because the original Matlab code uses values only
from the on-design model (which means that is needed to run it (10)
only one time). A


The first step in the calculation of the “expected” effectiveness
consists of the “Dittus and Boelter” correlation [11] (with the Calculation of the “B” constant: the B constant is
conditions of forced convection inside pipes, turbulent flow, the way for calculating the expected effectiveness.
cooled fluid), with the objective to define the Nusselt number It comes from the comparison between the Nusselt
of the hot exhaust at the inlet of each heat exchanger. The number at the reference conditions and the Nusselt
expected performance is calculated assuming that heat number calculated at the operating conditions.
resistance on water/steam side is negligible with respect to the
UREF ⋅ D ⎛ ρREF cREF D ⎞ ⎛ μREF c pREF ⎞
⎜ ⎟
hot flue gas side. Such a correlation is reported in (6):

0.3

⎜ ⋅ ⋅ ⎟
0.8

λ 0.023 ⎜ ρ c D⎟ ⎜ μ cp ⎟
⎛ ρ ⋅c⋅ D⎞ ⎛ μ ⋅ cp ⎞ = REF = ⋅ ⋅⎜ ⎟
U ⋅D U⋅D 0.023 ⎜ μREF ⎟ λREF
NuREF
Nu = = 0.023 ⋅ Re ⋅ Pr = 0.023 ⋅ ⎜⎜ ⎟⎟ ⋅ ⎜⎜ ⎟ ⎜ ⎟
0.8 0.3
(11)
λ ⎟ ⎜ ⎟
⎝ μ ⎠ ⎝ λ ⎠ λ μ ⎜ λ ⎟
0.8 0.3 Nu
⎝ ⎠
(6)
⎝ ⎠

Using for each heat exchanger the procedure explained below, Ordering the equation and separating the known terms from the
the “expected” effectiveness can then be calculated, using the unknown terms, (12) can be obtained, describing the value of
thermal transmittance of the heat exchanger, the number of the heat exchanger transmittance U at the current operating
thermal units and the characteristic parameter R. conditions.

• ⎛ ρ ⋅ c ⎞ ⎛ μ ⋅ cp ⎞
= B ⋅ λ ⋅ ⎜⎜ ⎟⎟ ⋅ ⎜⎜ ⎟⎟
0.8 0.3
Calculation of the reference effectiveness (εREF)

⎝ μ ⎠ ⎝ λ ⎠
using the thermodynamic data coming from the U EXP (12)
Gate Cycle® model, with the following ambient
conditions (called “reference conditions”) :
Tamb = 288 K , pamb = 1.013 bar , HR = 60 %. The B constant can be written as:

B=
Calculation of the reference characteristic ratio

ϕ
U REF
R*REF (7) using thermodynamic data coming from (13)
the Gate Cycle® model at the ambient conditions

Where the value of the ϕ constant:


described in the previous point

RREF * = min
E
⎛ρ ⋅c ⎞ ⎛ μ REF ⋅ c p REF ⎞
(7)
ϕ = λ REF ⎜⎜ REF REF ⎟⎟ ⋅⎜ ⎟
Emax 0. 8 0.3

⎝ μ ⎠ ⎜ λ REF ⎟
(14)
• Calculation of the reference number of thermal
REF ⎝ ⎠
units NTUREF of the heat exchanger, as inverse
formula coming from the one that represents the It is evident, from the calculation sequence presented before,

⎛ ε REF − 1 ⎞
heat exchanger effectiveness: that the calculation of the U value at the operating conditions is

ln⎜⎜ ⎟
independent from the coefficient “0.023” originally used in the

⎝ RREF * ⋅ε REF − 1 ⎟⎠
Dittus and Boelter correlation. This is very important, because

=−
1 − RREF *
(8) it provides the calculation scheme with a more general validity.
NTU REF

6 Copyright © 2008 by ASME


Once the U value at operating conditions (UEXP) is calculated, 1) Calculation of the partial derivative of the output of
the values of NTUEXP and R*EXP can be obtained from (15) and the system with respect to the error of one
(16): measurement

U⋅A
2) Partial derivative at point 1) times the measurement

NTU EXP =
uncertainty (supposed as 1% of the read value)
(15) 3) Divide the product at point 2) with one of the output
E min

⎛ ∂f Δx SM i ⎞
Mathematically is expressed by (19):

R *EXP = smi , j = ⎜ j ⋅ ⎟ ⋅100


Emin
⎜ ∂x ⎟
⎝ ⎠
(16) (19)
Emax f
i j

Finally, the “expected” effectiveness of each heat exchanger at where:


the operating conditions can be calculated through (17): i = number of rows of the sensitivity matrix (i = 1, m)

1 − e − NTU EXP (1− R*EXP )


ε EXP =
j = number of columns of the sensitivity matrix (j = 1, n)

1 − R *EXP ⋅e − NTU EXP (1−R*EXP )


m = number of measurements
(17) n = number of outputs

The “sensitivity matrix” can be used for the so-called “matrix


As a final comment, an alternative index to NDPI of (5) could inversion” (a developing tool for estimating physical conditions
be defined using the ratio of the actual and on-design heat of the system in zero-degraded status), which is out of the
exchanger surface, as reported in (18). The “actual” heat scope of the present paper. The direct result of the compilation
exchanger surface is an input of the system, while the of the “sensitivity matrix” is the calculation of the “error
“expected” surface is calculated using a procedure similar to matrix” and, afterwards, of the Index of Expected Error (IEE).
the one used for calculating the expected value of the The “error matrix” is obtained multiplying each element of the
effectiveness. “sensitivity matrix” with the real measurement error of each
probe.
NDPI A = ⋅ 100
A ACT
⎛ ∂f Δx EM i ⎞
emi , j = smi , j ⋅ Δx EM i = ⎜ j ⋅ ⎟ ⋅100
(18)
A EXP
⎜ ∂x ⎟
⎝ ⎠
(20)
i f j
However, such a Performance Index is affected by a high IEE,
as demonstrated in the following sensitivity and error
propagation analysis. The substantial difference between the two matrices is the
value of Δx; in the “sensitivity matrix” it is set equal to 1% of
3.6 SENSITIVITY AND ERROR PROPAGATION the read value, while in the “error matrix” the value of Δx is
ANALYSIS related to the real measurement uncertainty. In Table 4 the
considered values of Δx are reported, which are estimated from
The aim of the sensitivity and error propagation analysis the plant.
is to define the accuracy of calculated values, underlining the
uncertainty of measurements influence on them. The index KIND OF MEASUREMENT UNCERTAINTY
which defines the reliability of the result is called IEE, Index of Gas/Exhaust Flow Δ = ± 2% of read value
Expected Error. The wish is to link the sensor error in the Water Flow Δ = ± 1% of read value
measurement with the error in the calculation of the outputs, Steam Flow Δ = ± 3% of read value
which are the exhaust temperatures and the NDPIs. Two Pressure Δ = ± 0.25% of 2/3 of full scale
distinct matrices are built: the “sensitivity matrix” and the Temperature < 523 K
“error matrix”. (thermoresistance) Δ= ± 1 K
The “sensitivity matrix” imposes an error of 1% of the read Temperature > 523 K
value on each probe, and it calculates the variation of each (thermocouples) Δ= ± 3 K
output of the code (efficiencies, NDPI), which comes, in
general, from more than one measurement. The procedure to Table 4: Measurements uncertainties
calculate the “sensitivity matrix” is the following:
The “error matrix” is created in order to calculate the values of
the IEE for each heat exchanger; such indexes are defined,
according to [9], as explained in (21).

7 Copyright © 2008 by ASME



this performance analysis are reported, considering one

IEE j =
m economizer (ECO2), one evaporator (EVA1) and one
i =1
(emi , j ) 2 (21) superheater (SH6). In the following graphs the number in the x
axis represents the number of the data acquisition: the
In Table 5 is possible to compare different columns of the operating days of the plant corresponding to the numbers in the
“error matrix” related to one day of functioning of the plant. graphs were taken in a time interval of about 2 years.
The columns are related to the NDPI and the NDPIA of one
superheater (SH5), one evaporator (EVA2) and one economizer
(ECO3) while the corresponding IEE is shown at the bottom of
the column.

NDPISH5 NDPIEVA2 NDPIECO3 NDPIASH5 NDPIAEVA2 NDPIAECO3


m1f -0.0942 -0.1275 -0.1042 -0.3541 -0.2304 -0.1781
m2 0.0001 0.0001 0.0001 0.0004 0.0002 0.0001
m3 0.173 0.2194 0.1552 0.0096 - 1.6094 0.7353 0.3487
m10 0.0144 0.0389 0.0411 0.0374 0.0876 0.0784
m16 0 0.0016 0.0012 0 0.0035 0.0023
m19 0.0004 0.0005 0.0004 0.0026 0.0011 0.0007
T1f -0.0105 -0.013 -0.0098 -0.0627 -0.0272 -0.0183
Tamb 0 0 0 0 0 0
T0 0.0127 0.0158 0.0118 0.0898 0.0343 0.0224
T2 0 0 0 -0.0001 0 0
T4 0 0 0 0 0 0
T5 0 0 0 0.0481 0 0
T6 -0.0078 -0.0096 -0.0072 -0.0052 -0.0201 -0.0134
T7 0 0 0 0 0 0
T8 -0.0146 -0.018 -0.0136 -0.0846 -0.0373 -0.0252
T9 0.0141 0.0175 0.0131 0.1005 0.0381 0.0248
T10 0.0023 0.0028 0.0021 0.0148 0.006 0.004
T11 0.0553 0 0 -0.0158 0 0
T12 -0.011 0.0639 0 0 0.1503 0
T13 0 -0.0024 0.0204 0 -0.0051 0.039 Figure 4. NDPI of ECO2 (Economizer)
T14 0 0 -0.0068 0 0 -0.0126
T15 0 0.0015 0.0011 0 0.0032 0.0021
T16 0 -0.0006 -0.0004 0 -0.0012 -0.0008
T17 0 0 0 0 0 0
T18 0 0 0 0 0 0
T20 0 0 0 0 0 0
pf 0 0 0 0 0 0
p0 -0.0007 -0.0008 -0.0006 -0.0044 -0.0018 -0.0012
p1 0 0 0 0 0 0
p2 0 0 0 0 0 0
p3 0 0 0 0 0 0
p4 0 0 0 0 0 0
p5 0 0 0 0.0012 0 0
p6 0 0 0 0 0 0
p7 0 0 0 0 0 0
p8 0.0005 0.0006 0.0005 0.0033 0.0014 0.0009
p9 -0.0002 -0.0002 -0.0002 -0.0012 -0.0005 -0.0003
p10 -0.0001 -0.0001 -0.0001 -0.0005 -0.0002 -0.0001
p11 -0.0005 0 0 0.0008 0 0
p12 0.0007 -0.0001 0 0 -0.0002 0
p13 0 0 0 0 0 0
p14 0 0 0 0 0 0
p15 0 0 0 0 0 0
p16 0 0 0 0 0.0001 0.0001
p17 0 0 0 0 0 0
p18 0 0 0 0 0 0
p20 0 0 0 0 0 0

IEE 0.1817 0.2289 0.1897 0.6045 0.4779 0.3335


Figure 5. NDPI of EVA1 (Evaporator)
Table 5: Error Matrix Columns

It is evident that the IEE related to the NDPIA assume higher


values with respect to the IEE related to the NDPI, for this
reason it was decided to use only the NDPI instead of both
NDPI and NDPIA.

4 ANALYSIS OF HRSG HISTORICAL DATA

The calculation procedure illustrated before was applied


to the test set; in the next figures the results (NDPI and IEE) of

8 Copyright © 2008 by ASME


the plant. TPG is developing this kind of analysis, in order to
evaluate in terms of “fuel impact” [5] [6] the degradation of the
heat exchangers. A fundamental element in this calculation is
represented by the procedure called “matrix inversion”,
because it permits the calculation of HRSG performance
considering the heat exchangers as zero-degraded. The results
coming from this calculation will be presented in the next
papers.

5 CONCLUSIONS

A complete procedure for calculating HRSG performance


of large size combined cycle is explained in this paper.

A detailed description of the original Matlab code is made


and the results obtained are presented.
Figure 6. NDPI of SH6 (Superheater)
In each figure the straight blue line represents the instant The method was applied to an existing plant, using data
calculated value of the NDPI, while the two purple lines coming directly from the monitoring system. The procedure
represent the instant value of the IEE and its use as the inferior shows good capabilities in estimating the degradation of the
and superior limit for defining the reliability of the results. HRSG throughout plant life.
The straight red line represents the trend of NDPI; this line The main features of the monitoring/diagnostic approach
clearly depicts a degradation in all the heat exchangers of the described here are:
HRSG.
The highly oscillating trend of the instantaneous value of the • possibility of observing the instantaneous
NDPI is caused mainly by two factors: uncertainty of each

performance of each heat exchanger;
measurement and fluctuation of the physical parameters. The
assessment of accuracy of each performance index at
reliability of the results coming from the calculation is

the current operating condition;
instantly verified by the evaluation of the corresponding IEE
possibility to improve plant availability and
for each heat exchanger. This tool can be used to calculate the
profitability through the continuous update of
instantaneous performance of each heat exchangers on-line,
scheduled maintenance (event-driven rather than
with a delta-time of less than 5 minutes. It is also possible to

time-driven maintenance);
calculate the NDPIs off-line, for tracing trend lines of heat
possibility to extend the interval between two stops
exchangers performance weekly, monthly or yearly, in order to
suggest maintenance intervention on heat exchangers and for of the plant for maintenance reasons, focusing on the


results of the analysis of historical logged data;
delayed maintenance stops.
troubleshooting and early warning of malfunctions
From the previous graphs is possible to observe a and dysfunctions [5] [6].
consistent negative trend in most of the heat exchangers NDPI.
Considering a maximum time interval of around two years is Work in progress and future progresses of this work are:
possible to observe that the heat exchangers performance is
getting worse. 1. development of a steam turbine performance
Evaporators seem to be more degraded (from 5% to 20%): this monitoring system;
could be logical, because of the fact that in each evaporator 2. development of a steam condenser performance
there is a consistent pressure drop. Economizers and monitoring system;
superheaters seem to be less degraded than the evaporators, in 3. troubleshooting-oriented model for the whole plant;
fact their degradation trend is around 5% - 10%. 4. thermo-economic analysis of the whole combined
It is important to observe that the instantaneous value of the cycle power plant [7].
NDPI (straight blue line) is placed in the interval defined by
upper and lower limit, fixed by the IEE. This means that the 6 ACKNOWLEDGEMENTS
results are reliable.
The authors wish to acknowledge Ansaldo Energia for the
Local and global influence of heat exchangers degradation partial support to the present research activity.
depicted in these graphs will be verified and quantified through
the development of a complete thermo-economic analysis of

9 Copyright © 2008 by ASME


REFERENCES

[1] “Gas Turbine Condition Monitoring and Fault


Diagnosis”, von Karman Institute for Fluid Dynamics,
January 13-17, 2003

[2] A. I. Zwebek, P. Pilidis, 2001, “Degradation Effects on


Combined cycle Power Plant Performance, Part 1: Gas
Turbine Cycle Component Degradation Effects”. ASME
Paper 2001-GT-388.

[3] A. I. Zwebek, P. Pilidis, 2001, “Degradation Effects on


Combined cycle Power Plant Performance, Part 2: Steam
Turbine Cycle Component Degradation Effects”. ASME
Paper 2001-GT-389.

[4] A. I. Zwebek, P. Pilidis, 2001, “Degradation Effects on


Combined cycle Power Plant Performance, Part 3: Gas
and Steam Turbine Degradation Effects”. ASME Paper
GT-2002-30513.

[5] V. Verda, L. Serra, and A. Valero, 2002,


“Thermoeconomic diagnosis: Zooming Strategy Applied
to Highly Complex Energy Systems. Part 1: Detection and
Localization of Anomalies”, ASME IMECE, New
Orleans.

[6] V. Verda, L. Serra, and A. Valero, 2002,,


“Thermoeconomic diagnosis: Zooming Strategy Applied
to Highly Complex Energy Systems. Part 2: On the
Choice of the Productive Structure” ASME IMECE, New
Orleans.

[7] S. Uson, L. Correas, A. Valero, 2006, “Anamnesis for


improving thermoeconomic diagnosis. The case of a
3x350 MW coal-fired power plant”. ESDA2006-95341.

[8] R.R. Gay, C. A. Palmer, M. R. Erbes, 2004, “Power Plant


Performance Monitoring”, R-Squared Publishing, ISBN
0-9755876-0-9.

[9] ASME 19.1 – 1985, Measurement Uncertainty


(Instruments and Apparatus).

[10] D. Therkorn, “Remote monitoring & diagnostic for


combined-cycle power plant”. ASME Paper GT-2005-
68710.

[11] S. Kakac, “Boilers, evaporators & condensers”, Wiley-


Interscience.

10 Copyright © 2008 by ASME

View publication stats

You might also like