WO2005090764A1 - Method and device for diagnosing a turbine plant - Google Patents

Abstract

The invention relates to a method and device for diagnosing a turbine plant (6) comprising a gas turbine (8) having a number of components (10, 12). An additional output is automatically predicted by which an operating output of the gas turbine (8) is increased when eliminating a soiling of one of the components (10, 12).

Description

description

Method and device for diagnosing a turbine system

The invention relates to a method and a device for diagnosing a turbine system with a gas turbine comprising several components, the gas turbine having at least one compressor and one filter.

With increasing runtime of a gas turbine, its performance and efficiency decrease due to the influence of aging and pollution. The aerodynamic parts of the gas turbine compressor are particularly affected. The entire power plant process is adversely affected by aging. The aging mechanisms are pollution, deposition, erosion and corrosion. Contamination and deposits are temporary properties that can be reduced or eliminated by so-called online and offline washes. Online washes can be carried out during operation and must be carried out at least once a day. Offline washes require the turbine system to be down for about six hours; they will usually be done once a month.

Contamination of the compressor is caused by particles adhering to the surfaces. Oil and water mist help dust and aerosols to stick to the blades. The most frequently occurring soiling and deposits are mixtures of water wetting, water-soluble and water-insoluble materials. Contamination from ash deposits and unburned, solid cleaning products can occur in the gas turbine. Such air pollutants adhere to the components of the flow path of the gas turbine like scales and react with them. The loss of material due to chemical reactions of metals with pollutants is called hot corrosion. Beat larger, hard particles that are larger than 20 µm, on the surfaces of the flow components, material abrasion occurs.

Another aspect for the aging of the turbine system is the abrasion of components. Here surfaces of rotating and stationary components rub against each other. These friction effects often occur when the system is started up because the materials expand unevenly due to the temperature increase. Many components were realized with surfaces that allow a certain abrasion. Abrasions by

Particle impact and abrasion are commonly referred to as erosion.

In addition, pieces of ice that have formed at the inlet of the gas turbine can loosen and hit the flow components. To prevent this, a so-called anti-icing system is used. Here, air preheating prevents the temperature of the air from dropping below freezing when entering the gas turbine and thus preventing the water from freezing. An anti-icing (English ^w anti-icing ") of the gas turbine is usually carried out from a temperature of 278.15 K.

Signs of aging, such as deposits and dirt, can be improved by cleaning. On the other hand, for components that are heavily eroded, corroded and worn out, only a repair or replacement of the component helps. Tests have shown that the decrease in performance due to pollution stabilizes at around 90%. Particles are released from the growing layers of dirt, which migrate through the turbine and kick off other contaminants. The following therefore applies to the pollution _factor f pollution:

"- J pollution -" ^• 1 L 1]

The contamination and erosion causes an increased surface roughness of the blades. this leads to large friction losses in the gas turbine. The laminar boundary layer flow can change into a turbulent flow and there is an increase in losses due to the increasing flow resistance. During the aging process of the gas turbine, the radial gap increases due to abrasion and corrosion. The gap flow increases and the performance of the system decreases.

Geometric changes in the blades, due to erosion, deposits and damage, result in reduced performance of the gas turbine. Deposits, erosion and corrosion at the entry lead to changed entry angles, which have a very strong effect on the thermal performance. The influence of blade aging is particularly great at the gas turbine inlet. An aged compressor leads to a reduced distance from the surge line and the stall. It is characterized by lower performance and additional losses.

The aging of the compressor has a negative impact on the gas turbine efficiency η _GT , the gas turbine power P _GT and the gas turbine outlet mass flow m ^ l. Positive effects for the circuit in a turbine plant, e.g. B. in a gas and steam power plant, only result from a slightly higher outlet temperature and better steam turbine efficiency. However, the predominant factor here is the significant reduction in the outlet mass flow, which has a direct influence on the achievable performance. Studies show that 5% compressor aging leads to a 8% reduction in gas turbine performance and a 3% reduction in gas turbine efficiency. For the entire gas and steam power plant, it leads to a reduction of approx. 1.5% power and 0.5% efficiency.

In order to counteract the reduction in performance of the turbine system, regular compressor washes are carried out. compaction Buckets can be washed in online and offline mode. In online mode, the turbine system continues to operate during cleaning, whereas in offline mode it is shut down completely and rotated for about six hours with a K-cell rotating device to cool it down.

Offline washing results in greater performance recovery than online washing. With the help of offline washing, performance recoveries of up to 3% can be achieved. Online washing results in an average recovery of approx. 1_% of power. The most effective window cleaning can be achieved with a combination of online and offline washes. Regular online washing extends the time intervals between the required offline washes.

The system must be shut down for offline washing. To avoid thermal stresses, it is cooled for six hours using a shaft rotating device. If offline washing represents a significant impairment of operation, the blades are washed online. The gas turbine load is only slightly reduced. Online washes are mainly used to avoid the build-up of the dirt layer. Online washing is usually carried out once a day with demineralized water and washing with detergents every third day. Offline washing should take place once a month or after a so-called "trip". If the turbine system has not been cleaned for more than 350 hours, offline washing must be carried out, since the online cleaning method can no longer remove dirt.

The appearance of aging is generally divided into the areas "regenerable aging", "non-regenerable aging" and "permanent losses", which are described in more detail in Table 1 below. Table 1 - Categorization of aging and pollution: Category Example Countermeasures permanent old surface roughness, none, as a rule component changes too expensive, non-rain-increasing gap-blade change, replaceable old luster, regenerable soiling, ^' drain online and / or

Pollution wrestled offline laundry

FIG. 1 for the prior art shows a basic course of the aging of a gas turbine based on a new and clean system condition. The history shows how the performance parameters decrease over the course of the operating time.

Permanent losses are caused by increased surface roughness and changes in components. The replacement of the affected components would not be worthwhile because the costs incurred outweigh the benefits achieved. The permanent signs of aging are shown in FIG. 1, section a. Phenomena that cannot be regenerated, such as increasing gap losses, increasing cooling air consumption and lower turbine efficiency, can be reduced by changing the turbine blades; see FIG. 1, section b.

The third phenomenon of aging is considered regenerable. The natural aging process can be postponed to a certain extent by changing the filter and washing offline; see FIG. 1, section c. Curve 1 of FIG. 1 shows the aging that could be achieved by hand washing, changing the filter and thermodynamically optimizing the gas turbine.

However, this is not feasible for the operator since he does not optimize the gas turbine thermodynamically carry out and hand washing is only possible after a long standstill. Therefore, the reset c cannot take place up to curve 1. The influence of online washes is hidden behind curve 4. It becomes clear that with every online wash a performance and heat rate recovery is achieved. An optimal system condition can only be achieved with regular offline and online washes. If this is neglected, the losses of the plant increase.

The gas turbine also includes an upstream air filter. A pressure loss takes place via this filter, which affects the thermal performance of the gas turbine and the overall system. This pressure loss increases in the course of operation. In addition, the pressure drop is dependent on the current volume flow, air humidity, temperature and pressure.

mitThis pressure loss is, for example, in a control and / or regulation program for the turbine system, for. B. in the so-called program KREISPR, used as an input variable. There is a conversion to ISO conditions. The corresponding equation is:

With

4PA, __ _{Ό Default} value total pressure loss (at ISO) ^ _. t / fy _/ o mass flows input R _AI , R _VI0 relative air humidity input P _O 'P _AI pressure input T _j u'Tγio temperatures input

The contamination of the gas turbine, in section 1 in section A prior to online washing, cannot be determined with certainty. The influences and results are too noisy. The optimal time for online laundry is currently through determines the operator based on purely economic aspects, e.g. B. in low load times. That is, the previous decisions about the time of removal of contamination of one of the components of the turbine system, e.g. B. by washing the compressor, e.g. B. an online or offline laundry, or by replacing the filter, are based only on empirical values under economic aspects or under preliminary studies with fixed boundary conditions.

The invention is therefore based on the object of specifying a method and a device for diagnosing a turbine system having a gas turbine comprising a plurality of components, the operation of the turbine system, which is as cost-effective as possible, being made possible at the same time.

According to the invention, the object with regard to the method is achieved by the features of claim 1. With regard to the device, the object is achieved according to the invention with the features of claim 24.

Advantageous further developments are the subject of the dependent claims.

The invention is based on the consideration that for a turbine system that is both low-wear and high-performance, it should be monitored and continuously diagnosed. For this purpose, in particular an additional power to be achieved by removing contamination of the turbine system is determined and continuously determined. In other words: In addition to the additional power to be achieved, the present diagnostic concept determines the current power loss due to a degree of contamination on which the turbine system is based, eg. B. by a dirty filter or a dirty compressor. In detail, the method for diagnosing the turbine system with a plurality of components, for. B. a compressor and a filter, comprehensive gas turbine automatically predicts the additional power by which an operating power of the gas turbine is increased if one of the components is removed from contamination.

For a simple decision on the type and scope of pollution removal, depending on the value of the determined additional service and its prognosticated course in the future, it is determined whether the gas turbine is temporarily shut down to remove the pollution. In addition, depending on the total effort of a temporary shutdown, a time for a temporary shutdown is determined. In determining the point in time for the temporary shutdown, in particular operating costs, for example electricity generation costs, the proceeds for electrical energy, costs for removing the pollution, e.g. B. washing costs or costs for a filter change are taken into account.

In one possible embodiment, the additional power is determined on the basis of at least one model calculation of the gas turbine and at least one operating variable representing the gas turbine. A static and / or a dynamic company variable is or is determined as the company variable. Time-dependent measured values, in particular ambient pressure, compressor outlet pressure, pressure loss, operating hours, time of last removal of contamination, compressor mass flow, compressor inlet temperature, compressor outlet temperature, instantaneous turbine output, are determined as the dynamic operating variable. The dynamic operating variables are therefore current or instantaneous measured values recorded on the gas turbine, for example by means of sensors.

Geometric dimensions of the gas turbine, duration of removal of the contamination, electricity generation costs and / or total output of the turbines are used as the static operating variable. Line system and / or the gas turbine and / or thermodynamic constants, in particular standard pressure, standard temperature, standard air humidity, determined. The static operating variables are characteristic output or operating variables that describe the gas turbine and its components in more detail.

For a sufficiently precise determination of the additional service on the basis of the model calculation, the company size is validated on the basis of at least one plausibility check and / or a stationary check, in particular on current environmental conditions. The validated farm size is then processed using the model calculation. A stationarity check of the dynamic operating variables or measured values is understood in particular to mean the determination of the deviation of the measured values, taking into account a tolerance range for the deviation of a current measured value from a previous measured value.

As an alternative or in addition, the measured values can be combined before processing, which brings about a reduction in the number of measured values to be processed by means of the model calculation. For example, the measured values are averaged over a period of 15 minutes. Non-plausible measured values identified on the basis of the plausibility check are not included in the time averaging. Measured values from faulty sensors are also not taken into account in the averaging. The averaged measured values are also checked for plausibility of their stationarity. The recorded measured values can also be used instead of validated measured values. This can make it a big one

Scattering of the results of the model calculation come. In addition, so-called network measured values can measure several measurements of the same measured variable, e.g. B. distributed several temperature measurements over a pipe circumference, spatially averaged very close. Non-plausible network measured values are discarded before spatial averaging or, alternatively, their confidence intervals are expanded. Based on the validated farm size, which are validated for current environmental conditions, e.g. B. at full load of the gas turbine, a current degree of pollution, a current grade and / or a current state of aging of one or more components of the gas turbine and / or an isentropic compressor efficiency for an unpurified and / or a contaminated gas turbine are determined by means of the model calculation. Additionally or alternatively, based on the validated operating variables by means of the model calculation, a first additional service, which can be achieved by eliminating contamination by offline washing of the gas turbine, in particular a compressor, and / or a second additional service, which is achieved by removing contamination of the Gas turbine can be reached by washing a compressor online, and / or a third additional service, which can be achieved by removing a contamination of the gas turbine by changing a filter, predicts.

In order to be able to compare the ascertained diagnostic parameters, ie the first additional service, the second additional service, the current degree of contamination, the current state of aging, the current level of quality and / or the compressor efficiency, the operating parameters are preferably standardized. For example, the operating parameters are standardized to reference conditions, in particular ISO conditions, and then processed using the model calculation. Using the model calculation, the standardized operating variables are used to determine a reference degree of pollution, a reference quality level and / or a reference aging state of one or more components of the gas turbine and / or an isentropic reference compressor efficiency for an unpurified and / or one contaminated gas turbine determined. Additionally or alternatively, a first reference additional service, a second reference additional service and / or an isentropic reference efficiency are based on the standardized operating variables (= in relation to reference ambient conditions) using the model calculation determined.

Depending on the specification, a loss of power and / or additional power validated to current environmental conditions and / or additional power in relation to reference environmental conditions can additionally or alternatively be determined on the basis of the standardized operating variable and the current validated operating variable using the model calculation.

For a quick and easy evaluation of the diagnosis of the gas turbine, the diagnostic variables determined on the basis of the model calculation are reduced. For example, a grade with a value greater than 1 and / or a negative degree of pollution are rejected. As a further condition for performing a diagnosis of the gas turbine, it can also be specified that the diagnosis is only carried out when the gas turbine is operated in the stationary full-load state and not with anti-icing. For a simple evaluation of the diagnosis, a predicted and / or determined additional service can also be rejected if the associated value is negative.

In a further embodiment, a diagnostic variable ascertained on the basis of the model calculation, in particular an additional power, a power loss, a degree of contamination, is output. For example, the diagnostic size determined is displayed visually. Depending on the specification, a regression function determined on the basis of the diagnostic size for the last removal of a contamination, in particular through an offline wash or an online wash, can also be output. The regression function is preferably output as a function of a user-defined threshold value with a color change. In this way, contamination of the gas turbine or one of its components that goes beyond the usual level can be identified particularly quickly and easily by a user, so that appropriate measures, such as carrying out washing or filter replacement, can be taken.

Alternatively or additionally, the regression function with a prognosis about the future course of the diagnostic variable, for example from a predetermined and thus elapsed period of time since the last removal of contamination, in particular through an offline wash, can be carried out independently of the determined diagnostic variable. B. with a color marking. In addition, when determining one of the diagnosis-large loss costs of a shutdown, washing costs, an additional profit after an offline wash and / or an amortization time of an offline wash can be taken into account.

In a possible embodiment for a device for carrying out the method, a forecast module is provided for determining the additional power, by which the operating power of the gas turbine is increased in the event that contamination of the gas turbine is removed. Depending on the type and design, the forecast module can be designed as a software module or as an electronic circuit. The forecast module is expediently implemented in an automation system for controlling and / or regulating the turbine system.

The advantages achieved by the invention are, in particular, that an operator of a turbine system uses the determined diagnostic size, in particular an additional service that can be achieved depending on the elimination of contamination, for one of the components of the turbine system to have a current and predicted information status about the state of the Turbine system and its components, in particular about the current degree of pollution and a related power loss, is available. Taking into account data on the total effort for eliminating contamination when determining the diagnostic size, maintenance or service is only carried out if it is both technically and economically lent is required. Long-term trends in pollution are identified by the diagnostic model. Based on the determined diagnostic values, the operator is given a decision-making aid for the time of a wash, e.g. B. an offline wash. There is a parameterizable, monetary assessment of the power loss or failure costs. Based on the model calculation, thermodynamic operating variables of the turbine system determined online or offline are taken into account. Economic data, such as cost data, are only considered statically. In the case of an offline diagnosis, the customer or user is offered a user interface in which cost data can be updated and predictions can be made about the optimal washing time (point of worthwhile online or offline washing).

Embodiments of the invention are explained in more detail with reference to a drawing. In it show:

1 schematically shows a basic course for the aging of a gas turbine,

2 shows schematically a display for outputting diagnostic variables of a turbine system,

3 schematically shows an example of the course of an operating variable, a measured value or a diagnostic variable,

4.1 to 4.4 schematically in table form the flow chart for a method for diagnosing a turbine system,

5 shows a schematic diagram of the course of the power loss of the turbine system as a function of the degree of contamination of the compressor, 6 schematically shows a diagram of the course of the total costs of removing pollution as a function of time,

7 schematically shows a diagram with a family of curves of different total cost profiles with different loads of the turbine system,

8 schematically shows a forecast module for carrying out the method for diagnosing a turbine system,

9 shows schematically a display for outputting the diagnostic variables determined by means of the forecast module, and

10 to 14 schematically, in table form, the measured values, operating variables, diagnostic variables, etc.

Corresponding parts are provided with the same reference symbols in all figures.

2 schematically shows a turbine system 6 with a gas turbine 8, a compressor 10 upstream of the gas turbine 8, which in turn is preceded by a filter 12. By means of the compressor 10, cleaned fresh air FL is compressed by means of the filter 12 and supplied as compressed air vL to a combustion chamber 14. Natural gas is supplied to the combustion chamber 14 as fuel B. The compressed air vL is heated in the combustion chamber 14 and fed to the gas turbine 8 for relaxation as heated air aL and discharged as combustion air VL. The gas turbine 8 is coupled via a shaft 16 to the compressor 10 and a generator 18 for generating electrical energy. FIG. 2 shows an show with the diagram for the turbine system 6 shown, with the individual components - gas turbine 8, compressor 10, filter 12, combustion chamber 14, generator 18 - associated operating variables BG, such as. B. compressor efficiency, gas turbine efficiency, filter pressure, and their current measured values MW, such as. B. 86.66%, 36.63% and 3.71 mbar, assigned sincä and output. The operating variables BG shown there are dynamic operating variables which are continuously recorded and updated online. FIG. 3 shows one of the recorded operating variables BG, e.g. B. the filter pressure, shown as a function of time.

2 shows the turbine system 6 with the ascertained and validated values for the individual operating variables BG as described above as an indication which is made available to a user of a control system and / or control system for the turbine system. In this case, further values of determined operating variables BG, e.g. B. dissipation factor and power losses are displayed. The display is not limited to the BG sizes shown. Optionally or additionally, the course over the time of a single or several operating variables can be displayed on a further display, as shown for example in FIG. 3.

Using the flow chart shown in FIGS. 4.1 to 4.3 in tabular form for carrying out the method for diagnosing the turbine system 6, which comprises a gas turbine 8 with a compressor 10 and a filter 12, the diagnostic method is described in more detail below.

The Diagnoseve_trfahren is based on the consideration of the relative influence of removing contamination of the turbine system 6, e.g. B. by changing the filter and / or washing the compressor 10, for. B. to determine an online or offline Verdiciter wash. The diag- nose with sufficient accuracy and as simple as possible. In addition, the diagnosis should be easy to integrate into an existing diagnosis program, whereby existing modules should be used.

The diagnosis and. In the present case, calculation is carried out for different operating states, in particular for stationary full load states, as shown in FIG. 4.4. Partial load conditions have an excessive noise band due to an uncertain, measurable guide vane position. The uncontrolled anti-icing operation also leads to non-measurable backflows, which prevent reliable validation and diagnosis. A diagnosis for a turbine system 6 operating in anti-icing mode is therefore generally not carried out.

The diagnosis can be carried out both offline and online. Online diagnosis has more simplifying assumptions, such as B. standard density, standard gas composition etc. In the offline diagnosis, the user can enter this data manually.

An online diagnosis is described in more detail below:

An online diagnosis is divided into five categories: preprocessing (see FIG. 4.1), gas turbine diagnosis (see FIG. 4.2), post-processing, in particular calculation of compressor parameters, calculation of a power loss or additional power (see FIG. 4.3) and output of the diagnostic variable, z. B. by visualization (see FIG 4.4). The process is fully automated in the online diagnosis. Except for preprocessing, these categories are started manually in the offline diagnosis. The specified steps relate to a gas turbine 8. Depending on the type and design of the turbine system 6, eg. B. with several gas turbines (not shown) or as a gas and steam turbine system (not closer shown), a diagnosis for another gas turbine is carried out analogously (in the same load cases).

During the diagnosis, two parameters are determined for the compressor 10: a current isentropic compressor efficiency ^ _comp kei ISO conditions and a pollution factor f _{η, pollution} • Furthermore, an additional power Pz is determined as a power loss, i.e. the possible performance gain through removal of pollution by a Laundry or a filter change is determined. The efficiency of the gas turbine 8 (= heat rate) or the overall efficiency of the turbine system 6 is not calculated.

The process concept is described below. 4.1 to 4.4 show an overview of the individual steps. The steps are numbered below according to the calculation logic, but they can also be carried out in a different order.

In FIG. 4.1, step 1, there is a plausibility check of the incoming measured values MW, which are recorded, for example, by a control and / or regulating system and are stored in a database for the diagnosis. The measured values MW are dynamic operating variables BG, e.g. B. filter pressure, ambient pressure, compressor discharge pressure, compressor mass flow, compressor inlet pressure etc. In step 2, a stationarity check of the input data or measured values MW is carried out every 30 s. A corresponding flag is set and the values are written to the database. In step 3, the measured values MW are averaged over a period of 15 minutes. Implausible measured values MW and faulty sensors are not included in the averaging. In step 4, the calculated mean values are checked again for plausibility of the stationarity.

In step 5, a spatial averaging of the network measured values according to VDI 2640 is carried out with a further plausibility test. In a further step, the input data can optionally be checked for plausibility, obviously incorrect values can be accepted with standard values, and the confidence intervals can be extended if necessary.

The gas turbine diagnosis is described in more detail in FIG. 4.2. In step 6, the overall circuit diagram of the turbine system 6 is validated in accordance with VDI 2048. Mass and energy conservation balances are used for the validation. The validation changes the measured values MW in order to maintain a consistent state with the system model.

nV.AMB ^rGTValidated measurement values MW ^V (with index V for validated, Komp for compressors, m for mass flow, GT for gas turbine, AMB for ambient conditions, T for temperature, P for power) are now available for the further calculation steps: • v ^m Comp

nV.AMB ^r GT

The following measured values MW, in particular pressure measured values p, are not validated:

where the superscript index VL denotes the use as an unvalidated measured value MW (= measured value).

Operating points, such as B. another gas turbine in the bypass or only one strand of a gas turbine (also called GT-HRSG for short) in operation are taken into account by "cycle segments" that can be switched off. The validated values are written into the database with the extension ".V". In step 7, as part of the validation calculation, an intensely compressor efficiency (with "Komp" = compressor) is determined: j isentrop _ T -V, AMB Komp, Aus Komp, Ein 'Ih.Komp ~, _ »L -> J Comp, Off Comp, On

where the inlet and outlet enthalpies of the compressor 10 (abbreviated to "Komp") are:

" ^• Comp.On ⁼ "'Comp.On * Comp, On> PAMB) h Comp.Off

determine. The determined values are written into the database with the ending ^W .V ". The calculation [5] - [6] can also be carried out in gas turbine solo mode.

In step 8, a forecast or model calculation (also "expected" calculation) of the overall circuit diagram, ie the turbine system 6, is carried out with the current ambient conditions with regard to the gas turbine 8 using a control and / or regulating program (the so-called KREISPR program). and a calibrated type file that shows the clean condition. The compressor or compressor efficiency _h, 'κ ^ _mp k ^e i - ambient conditions result from the prognosis. The values from the forecast or expected value calculation are written into the database with the ending ".E".

φ"^υ = 60Ψo Hierbei werden als Brennstoff B Standardbrennstoffe Erdgas und Heizöl nach IS06976 oder ANSI 133.6 angenommen. Aus der Referenzrechnung ergibt sich ein Referenz-Verdichter-Wir- kungsgrad η '™ bei Referenzbedingungen. Dieser Wert wird mit der Endung ".Rl" in die Datenbank geschrieben.In step 9, a forecast calculation with reference conditions (ISO) is carried out. The ISO conditions for the ambient temperature ύ _A - _m , the ambient pressure p _mB and the ambient air humidity φ _mB are: ϊr ψso = i5 ° c

φ " ^υ = 60Ψo Standard B natural fuels and heating oil according to IS06976 or ANSI 133.6 are assumed as fuel B here. The reference calculation results in a reference compressor efficiency η '™ under reference conditions. This value is written to the database with the ending ".Rl".

In step 10, a check is carried out for the full load state of the turbine system 6. For this purpose, a binary signal from the control and / or regulation system, e.g. B. the so-called OTC controller used. Additional measuring points can also be used for a plausibility test. Active unregulated anti-icing operation leads to the exclusion of the invoice. Preheating the air causes losses in performance and efficiency. These are generally not corrected because the temperature increase in the air and the air mass flow (in the uncontrolled case) are unknown. If necessary, however, they can be taken into account.

In step 11, the current (dirty) compressor efficiency is calculated under ISO conditions. The same relationships are assumed here: "R SO _n V, lSO _ _j - .VV..AAMMBB '' ιlhh, .KKoommpp -" - lh, Komp 'I hh..KKoommpp "E E., ΛAMΛffiB •• ^{ö ■} "'fh.Komp

This value is written to the database with the ending ".R". The relationship between the reference compressor efficiency and the current efficiency under ISO conditions can be: v, ιso ₌ (ι_ f _ f \ _n R.ιso, _g , lh, Komp \ Jη, aging η, pollution) I, Comp ι. -> i

form. The factor f _nMtemng takes into account non-regenerable and permanent aging. In principle, it is a function of the equivalent operating hours: Jη, aging ~ Jη, aging VEOH) [10]

However, it can also be simply stored as a constant in the database. The aging condition or factor refers to the type file used. If a new type file is calibrated and used, the aging factor must be reset to zero.

A pollution factor for the compressor 10 is determined as follows:

Alternatively or additionally, a quality grade for the compressor 10 can also be determined, according to:

The upper limit can be checked with a data series. A plausibility check is carried out on the basis of the range of the pollution factor for pollution according to equation [1]. If values <0 for the pollution factor JV _{ersc mutzung} ^or values> are determined 1 for grade G _{h omp} the compaction ^¬ ters 10 (eg just after an off-laundering in the noise band.), They are not for a prognosis or diagnosis considered in the form of a trend line.

In step 12 according to FIG. 4.3, a power loss is determined in the postprocessing on the basis of the difference between the current turbine state and the clean state. For this purpose, a reference calculation (ISO conditions) of the gas turbine circuit diagram with the current pressure loss Δp _{pilter is determined} via the filter 12 under ISO conditions and stored in the database as ".M". To save an additional invoice, this conversion can also be carried out manually. According to equation [2], a measured value conversion based on the ISO condition can also be derived: "M, ISO _ A,. m Komp ^R AM AMB PlSO P filter = ^A P filter [13] ^m Komp) ^R ιso T _IS0 p _mB

The current measured pressure drop Ap _{Fj [ter} through the filter 12, converted to ISO conditions as Δp ^ °, is used as an input variable in the gas turbine forecast calculation.

A first additional power £ ± P ™ ^{temchange is determined} under reference conditions, which results from a filter change, according to: _Λ pFilter change _ pR SO _ pAp £ ° r -i Λ -I i ^Λi Qj. - ^r GT GT! • ■ *

In step 13, the validated gas turbine power with dirty filter 12 and compressor 10 is determined from the values of the reference calculation and the forecast calculation (also called "expected" calculation) based on ISO conditions, in accordance with:

A second additional power AP ^ F '"can be determined from equations [14] and [15] under reference conditions that result from offline washing:

A pofflineW _J1 _ f _ f \ (pR.ISO _ Λ pFilterwechsel _ pV SO \ MSI ^λr Cr ~ ^l JP, offline JP, aging) \ ^r GT ^Djr GT ^r GT) LOJ

The constant factor f _{P <offline} takes into account that the complete power loss cannot be recovered by an offline wash. The factor f _PιA ι, _augmentation is first assumed to be constant, for it may optionally include a trend line as a function of the equivalent operating hours are assumed in accordance with: JP.alternation ~ J P.alternation EOH) [17]

with t _E0H = equivalent operating hours since the last revision.

_{≥ Q} [ 18 ]Before the determined values are written to the database, a plausibility check is carried out, according to:

_{≥ Q} [18]

The determined values are written into the database with the extension ".E". On the basis of online measurements, individual measured values MW and / or values derived or validated from them, e.g. B. the power loss can be written into the database.

According to FIG. 4.3, steps 16, 17, a regression curve can additionally be displayed for an output, in particular display of the trend line or forecast line for a diagnostic variable and / or operating variable, for example according to FIG. 3. This regression curve is expediently reset at the time of a last, previous offline wash. The time of the last offline wash can be determined by a binary signal or alternatively by the wash speed, which must run for at least three hours.

There are two options for determining the regression curve:

First, a linear regression curve: A linear regression curve is formed from the number n of values for the current power loss that has been determined since the last offline compressor wash signal. The confidence interval is not displayed. Trend lines with a positive slope are discarded. On the other hand, an exponential regression curve: An exponential regression curve is formed from the number n of values for the power loss determined since the last offline compressor wash signal. The confidence interval is not displayed. A strictly monotonically falling curve of the compressor efficiency is assumed. If this is not the case, the system switches to linear regression.

In the display of the course of the measured values MW, the regression function, for. B. the linear regression, are output. The display can optionally be done with a color change (with user-defined threshold). Alternatively, a color change can also be displayed based on operating experience values, for example after four weeks without offline washing.

An offline diagnosis of the turbine system 6 is described in more detail below. There are two cases of offline diagnosis:

• Offline determination of individual load points of the turbine system 6, • Offline forecast or forecast of a so-called "break-even" point of worthwhile offline washing and / or filter changes.

For the offline determination of individual load points, the stored values for the operating variables BG are taken into account in the database. As already described in more detail above under FIGS. 4.1 to 4.2 for the online diagnosis, these values are preprocessed, that is to say recorded and possibly validated. The user can start the turbine diagnosis, determination of the compressor parameters and the power losses manually. Here, the user can optionally carry out all steps, in particular determine all load cases and the associated values, or only parts, ie only values for individual components, eg. B. only the compressor 10 or the filter 12 or for both of them. Is z. If, for example, only a forecast or expected calculation is carried out, the necessary already validated values "" .V "are automatically read from the database. The resulting OffZLine results are not written into the online database, they are expediently written in an associated database for offline sizes.

With offline diagnosis, a modular calculation of individual components and their operating variables BG or states is thus made possible, the geometric design being deactivated. However, the user can have smudge factors, e.g. B. for the boiler. In the Off1-ine version, many influencing parameters, such as B. the fuel composition to be changed. Further evaluation options for marginal costs etc. are available.

und Zu- satzkosten gleich ist (=. der so genannte "break-<even"-Punkt) ermittelt wird.For the offline diagnosis or offline forecast of the "break-even" point of the offline laundry and / or the filter change, the diagnosis or forecast module includes an additional function. The online database with the currently recorded measured values MW and the resulting validated values is taken into account. The user can, for example, interactively enter current values for static and / or dynamic operating variables BG, e.g. B. current Kos endaten and electricity prices, is requested. The regression function is continued and determined from these values, the point in time at which the difference is obtained

and the additional cost is the same (=. the so-called "break- <even" point) is determined.

The determination of an optimal washing time for the compressor 10 is described below using the forecast module. The forecast module is used to identify and determine a long-term trend of pollution based on a diagnostic model. The aim is to provide the operator with a To provide decision support for the time of offline laundry.

As a prerequisite, the operating parameters BG, such as compressor parameters and power losses, are taken into account in the online diagnosis:

1. Compressor parameters: • Validated isentropic compressor efficiency under ISO conditions η ° _mp ; Compressor quality grade G ^ jj?

2. ^Power losses or additional services: • First additional service £ j> ™ ' ^change under reference conditions or under ambient conditions that result from a filter change; • second additional power P ^ ^meW under reference conditions or under ambient conditions that result from offline washing; • third additional service AP '" ^eW with reference conditions or with environmental conditions that result from offline washing.

When the values are displayed, there are several point clouds. As described above, these point clouds are preferably used to lay regression functions as regression curves or compensation curves. Extrapolation of the regression curves means that pollution will be predicted in the future.

(^f) bezeichnet. Die Parameter a2, al und aO werden dabei mit einer nichtlinearen Berechnungsmethode bestimmt. Die Leistungsverluste der Gasturbine 8 aufgrund eines veirschmutz- ten Verdichters 10 laufen auf einen maximalen Wert von ca. 90 % hinaus.5 shows an example of such a time profile for the power loss P _G τ as a function of the degree of contamination of the compressor 10. An exponential decay curve with a constant term of the form: ΔP «(t) = fl ₂ exp (α ₁ t) + α ₀ Assume that the values on the compensation curve are included

( ^f ). The parameters a2, al and aO are determined using a non-linear calculation method. The power losses of the gas turbine 8 due to a contaminated compressor 10 amount to a maximum value of approximately 90%.

Depending on the type and structure of the forecast module, the power loss P _G τ as shown in FIG. 5 can be determined currently and thus online. Alternatively or additionally, the power loss P _G τ can also be displayed for the past from stored measured values MW. This means that both a diagnosis for the past and a forecast for the future and / or a trend can be determined from previous and currently recorded measured values MW.

In the case of an online diagnosis, the current state of the turbine system 6 is always taken into account. If the turbine system 6 tends to run with a low load, the curve below U is constantly stretched, i. H. the same loss of performance occurs later.

In order to take economic parameters into account in addition to technical parameters, such as the dynamic and / or static operating variables BG, for example, the respective effort associated with offline washing, in particular the amount of time, the cost, is taken into account. Offline washing should preferably be in a period of low electricity prices with a subsequent (longer) period of higher electricity prices. In order to make a detailed statement about the optimal time for an offline wash, a lot of assumptions would have to be made and these would always be updated by the user. In the following, constant revenues and costs for a generated _Megawatt-hour of electrical energy are assumed. The determination of the optimum washing time is described below using an exemplary turbine system 6, in particular a gas and Steam turbine power plant with. with a capacity of 200 MW.

The following diagnosis calculation relates to the optimal time for an offline wash for the compressor 10. The process steps apply analogously to the filter change and / or the soot blowing as a further possibility for removing the contamination in the turbine system 6. The specific observation period is:

^ = 24h [19]

According to the power loss P _G τ shown in FIG. 5, loss costs per day result in the amount of:

K ^ (t) = - (e _d -k _Λ ) äPgg ^r (t) t _ψΛ [20]

where e _{el is} the revenue for electrical energy on the market and k _{el are} the electricity generation costs (= operating costs) in EUR / MWh. The values e _el , k _el and t are parameterizable constants.

The current loss costs can be summarized, e.g. B, convert arithmetically and specifically per past day: τ ez {) ^[ 21 ^]

The cost of a wash is made up of the reduced profit during the wash (6 hours) and the costs. for staff and detergents: τrwaesche _ r, I ι_ \ ptot _f , τrWaschmittel r 99. Λ-GUD ~ ^K / a_ ( ^e el ~ ^K el) ^r GVD ^τ dur ~ ^{t JL Δ Δ J}

The factor k _last takes into account the load for the turbine system 6 which is possible at the time of washing and which can be implemented on the market. The larger the factor k _last , the lower the costs of a see. At a wash period of t ^ ^chen = 6 hours a total power P ^l Glund the turbine plant _j 6 200 MW, a factors tor k _IAST = l (possible full load) and detergent cost κ ^{Waschmi "el} of 2.000, - EUR arise total washing costs κ ^wash from approx. 74,000 EUR per wash.

The washing costs are also converted specifically for one day: ', v-washing washing ^ r - _J 1 ^K special ~ washing * * ^■ "

where J ^{wash represents} the periodicity of washing. If the operator (in extreme cases) were to wash every day, he would have specific costs of EUR 74,000 per day. If he washes every 30 days, however, he would have specific costs of only approx. EUR 2,500 per day.

The specific costs of the loss of power k ^ '" _pez and the washing k _p ^hen can be summed up to the specific total costs k ° ^amt : 7 total loss. 1 wash r _l . - ^k GT, spec ^{+ k} ' spec LJ

The optimum is the minimization of the total costs:

6 illustrates the connection: The curve shown as a solid line indicates the increasing loss costs, the bold-dotted curve represents the decreasing specific washing costs. The thin-dotted curve adds up these costs and represents the total effort. Your minimum is drawn with a circle. In the example shown in FIG. 6, the factor k _hst = 1 was assumed. At such high

The optimal washing time would be given after or every 54 days. The specific total cost history with a family of curves of

Total cost profiles with different loads is shown in more detail in FIG. 7. Its optimum shifts to shorter washing periods when the _load factor k is less accepted, is that expected low default costs during the wash. However, the cost of the specific loss of performance remains the same. The assumption here is that the additional power (e.g. approx. 5 MW) can be implemented on the market after washing.

7 also shows the influence of the possible power failure from 50% load to 100% load. It turns out that even with a possible half load it is worth washing about every 30 days. The extreme case with a factor k _hst = 0 (eg _standstill due to operational disruption or revision) leads to the lowest washing costs and washing is worthwhile even with low soiling. FIG 7 also clarifies that the optimum is very flat in all cases. The operator is thus shown a possible, in particular optimal, time window in which offline washing is recommended. A sharp or precise recommendation is avoided.

In addition to the above calculations, the additional profit per day after an offline wash can be determined and output:

ZG ^w = 24 (e _el -k _el ) AP ° ^neW [26]

The amortization time can be determined from this:

7G ^W AZ ^W == - [27] GA ^W

This is the point in time from which the effort for offline washing of the compressor 10 has been amortized again by the performance gain (assumed as a constant). As described in more detail above in relation to FIGS. 4.1 to 4.4, the forecast module PM, which is designed as a software module, is composed of several method steps which can be combined into several software modules, at least in two modules B1, B2, as shown in more detail in FIG :

In the exemplary embodiment according to FIG. 8, the two modules B1 and B2 serve the following functions: • The first module B1 (referred to as "TDYgt_thermo" by way of example) is used to carry out a thermodynamic analysis of the turbine system 6. The method steps are specified in more detail in the description of FIGS. 4.1 to 4.2. The first module B1 is implemented, for example, in C ++ and can be called up from a higher-level control and / or regulation module for the turbine system 6 or its diagnosis. Given the full load condition and validation, etc., the forecast module PM is started. All required constants, calculation values etc. are stored in the database (referred to as "TDY database" in the example). The tables shown in FIGS. 10 to 14 provide an overview of the necessary operating parameters BG for diagnosing the turbine system 6. The second component B2 (in the exemplary embodiment referred to as "TDYgt_finance") takes into account the effort for removing the contamination of the turbine system 6, as specified with reference to FIG 4.3. In addition to a time expenditure, a financial expenditure, as already described above, is taken into account and output. To take the effort into account, the diagnosis of the turbine system 6 can be carried out in offline or online mode. The second component is expediently implemented as a JAVA applet. FIG. 8 shows an example of a forecast module PM formed from a number of building blocks B1 to Bn. The forecast module PM includes the first component B1 for thermodynamic diagnosis of the turbine system 6 and the second component B2 for taking into account the effort for removing contamination of the turbine system 6, in particular according to the processing and diagnosis described in FIG. 4.3. The first and second building blocks B1 and B2 have a third building block B3 for preprocessing operating sizes BG and their measured values MW according to FIG. 4.1 and a fourth module B4 for validating the operating sizes BG and their measured values MW according to FIG 4.2 upstream. All of the blocks B1 to B4 of the forecast module PM access the data stored in a first database DB1, such as current and / or previous measured values MW and / or operating variables BG. In addition, a further database contains DB2 diagnostic models for existing turbine systems or standard turbine systems, to which the first module B1 and the fourth module B4 have access, for example for referencing standard conditions or for comparing the current diagnosis with a stored model.

Depending on the type and structure of the forecast module PM, this can also be implemented in a higher-level diagnostic system for the turbine system 6 of a control and / or regulating system. The PM forecast module or its implementation is subjected to a review process and is subject to version control.

To output the determined diagnostic values, e.g. B. the first and second additional service, the degree of pollution, the level of quality, these are stored in the database DBl. The actual output, e.g. B. via a display, as shown for example in FIG. 9, therefore does not carry out its own calculations, eg. B. Formation of the regression curves. The regression curves are, for example, in the first block B1, ie in the thermodynamic diagnostic part, determined. The regression curves must always be started at the time of the last offline wash.

The second component B2 then outputs, in particular displays, the determined measured values MW, the operating variables BG and the diagnostic variables, such as the first and second additional power, the degree of pollution, the state of aging, the compressor efficiency and / or the quality, whereby In addition to FIG. 5, trend lines of the calculated thermodynamic process or operating variables BG or diagnostic variables are also displayed.

In addition, the curves shown in FIGS. 6 and 7 can be determined online and continuously updated in the display.

Furthermore, the following variants are possible in the output and can be combined as desired:

• Display and determination of the family of curves shown in FIG. 7 with newly defined cost parameters; • Display and determination of the family of curves shown in FIG. 7 over the interval of the load factor; • takeover, d. H. Storage and / or updating of the newly defined cost parameters in the database DB1; • Update the display values as soon as new data, ie. H. new measured values MW and / or company sizes BG are available.

In summary, FIGS. 10 to 14 schematically show in table form the measured values, operating variables, diagnostic variables, etc. detailed.

Claims

claims 1. A method for diagnosing a turbine system (6) with a gas turbine (8) comprising a plurality of components (10, 12), an additional power being automatically predicted by which an operating power of the gas turbine (8) in the event of contamination being removed one of the components (10, 12) is increased. 2. The method according to claim 1, wherein it is determined as a function of the value of the determined additional power whether the gas turbine (8) is temporarily stopped to remove the contamination. 3. The method according to claim 1 or 2, wherein depending on the total effort of a temporary shutdown, a time for a temporary shutdown is determined. 4. The method according to claim 1, wherein the additional power is determined on the basis of at least one model calculation of the gas turbine (8) and at least one operating variable (BG) representing the gas turbine (8). 5. The method according to claim 4, wherein a static and / or a dynamic operating variable are or will be determined as the operating variable (BG). 6. The method according to claim 5, wherein the dynamic operating variable is time-dependent measured values (MW), in particular ambient pressure, compressor outlet pressure, pressure loss, operating hours, time of last removal of contamination, compressor mass flow, compressor inlet temperature, compressor outlet temperature, instantaneous Turbine power can be determined. 7. The method according to claim 5 or 6, wherein as the static operating variable geometric dimensions of the gas turbine (8), Time for a removal of the pollution, electricity generation costs and / or total output of the turbine system (6) and / or the gas turbine (8) and / or thermodynamic constants, in particular standard pressure, standard temperature, standard air humidity, are determined or will be , 8th . Method according to one of claims 4 to 7, wherein the company size (BG) is validated using at least one plausibility check and / or a stationarity check and the validated company size is processed using the model calculation. 9. The method of claim 8, wherein based on the validated operating variable using the model calculation, a current degree of contamination, a current grade of quality and / or a current aging condition of one or more components of the gas turbine and / or an isentropic compressor efficiency for an uncontaminated and / or a contaminated gas turbine (8) can be determined. 10. The method according to claim 8 or 9, wherein on the basis of the validated operating variables by means of the model calculation, a first additional service which can be achieved by removing contamination in the gas turbine (8), in particular by offline washing of a compressor (10), and / or a second additional service, which can be achieved by eliminating contamination of the gas turbine (8) by online washing of a compressor, and / or a third additional service, which is achieved by removing contamination of the gas turbine (8) by changing a filter (12 ) can be reached, will be or will be forecast. 11. The method according to any one of claims 4 to 10, wherein the operating variable is standardized and the normalized operating variable is processed using the model calculation. 12. The method according to claim 11, wherein the current degree of pollution !, the current state of aging, the current quality level and / or the compressor efficiency are determined on the basis of the standardized operating variables for reference conditions. 13. The method according to claim 11 or 12, wherein on the basis of the normalized operating variables by means of the model calculation, a reference degree of contamination, a reference quality grade and / or a reference condition of one or more components (10, 12) of the gas turbine (8) and / or an isentropic reference compressor efficiency for an uncontaminated and / or a contaminated gas turbine (8) is or will be determined. 14. The method according to any one of claims 11 to 13, wherein a first reference additional service, a second reference additional service and / or an isentropic reference efficiency are determined or will be based on the standardized operating variables by means of the model calculation. 15. The method according to any one of claims 11 to 14, wherein on the basis of the normalized operating variable and the current validated operating variable, a loss of power validated to current environmental conditions with respect to reference ambient conditions is determined by means of the model calculation. 16. The method according to any one of claims 11 to 15, wherein on the basis of the normalized operating variable and the validated operating variable using the model calculation, an additional service validated to current environmental conditions is determined in relation to reference ambient conditions. 17. The method according to any one of claims 9 to 16, wherein a grade greater than 1 and / or a negative degree of contamination are discarded. 18. The method according to any one of the preceding claims, wherein a diagnosis of the gas turbine (8) is carried out only when the gas turbine (8) is operated in the stationary full load state and not with anti-icing. 19. The method according to any one of the preceding claims, wherein a predicted and / or determined additional service is rejected if the associated value is negative. 20. The method according to any one of the preceding claims, wherein a diagnostic variable determined using the model calculation, in particular an additional service, a power loss, a degree of soiling, is output, in particular depicted visually and, depending on the specification, one using the diagnostic variable for the last removal of a contamination, in particular by an offline wash or an online wash, determined regression function is output. 21. The method according to claim 20, wherein the regression function is output as a function of a user-defined threshold value with a color change. 22. The method according to claim 21, wherein the regression function is output with a color change from a predetermined period of time since the last removal of soiling, in particular by offline washing, regardless of the determined diagnostic size. 23. The method according to any one of the preceding claims, wherein loss costs of a shutdown, washing costs, an additional profit after an offline wash or online laundry and / or an amortization time of an offline laundry or online laundry are determined. 24. Device for performing the method according to one of claims 1 to 23, wherein a forecast module (PM) is provided for determining an additional service, by which a Operating performance of a gas turbine (8) of a turbine system (6) is increased in the event that contamination of the gas turbine (8) is removed. 25. The apparatus of claim 24, wherein the forecast module (PM) is designed as a software module or as an electronic circuit. 26. The apparatus of claim 24 or 25, wherein the forecasting module (PM) is implemented in an automation system for controlling and / or regulating the turbine system (6).