EP4692562A1

EP4692562A1 - A method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly

Info

Publication number: EP4692562A1
Application number: EP24306325.2A
Authority: EP
Inventors: Joe HACHEM; Aymar DE BERGEYK
Original assignee: TotalEnergies Onetech SAS
Current assignee: TotalEnergies Onetech SAS
Priority date: 2024-08-05
Filing date: 2024-08-05
Publication date: 2026-02-11
Also published as: WO2026032939A1

Abstract

The invention relates to a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, the method comprising:
- measuring (110) the compressor inlet temperature and the compressor inlet pressure;
- receiving (120) a first measured value corresponding to the compressor inlet temperature, and a second measured value corresponding to the compressor inlet pressure;
- calculating (121) the compressor isentropic efficiency from the first and second measured values, using a predefined thermodynamic model;
- predicting (122) the compressor isentropic efficiency from the first and second measured values, using a predefined mathematical model;
- calculating (124) the real time difference between the calculated compressor isentropic efficiency and the predicted compressor isentropic efficiency;
- based on the calculated real time difference, providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator as to whether or not a compressor offline washing is necessary.

Description

TECHNICAL FIELD OF THE INVENTION

The present invention concerns a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly. The present invention also concerns an associated computer program product. The present invention also relates to an associated readable information carrier.

BACKGROUND OF THE INVENTION

On single-shaft gas turbine lines comprising a turbine-compressor assembly, it is known to proceed to an offline washing of the compressor when the turbine-compressor assembly is shut-down, in order to prevent the compressor from becoming too fouled. Currently, such a washing procedure is carried out at regular, empirically determined intervals, typically at a frequency comprised between one washing every 700 operating hours and one washing every 1200 operating hours, or approximately 5 to 9 washings per year. However, such repeated washings have a high economic cost. Besides, the optimization of such a procedure is not optimal, because the compressor is sometimes washed even though it is only slightly fouled. Last but not least, such repeated washings might ultimately degrade or even damage the compressor.

SUMMARY OF THE INVENTION

There exists a need for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, in a more cost effective and energy saving way, and which prevents too much degradation of the compressor by avoiding doing too many washings or not enough ones.
To this end, the invention relates to a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, the method being implemented by an assembly including a computer, at least one temperature sensor and at least one pressure sensor, said computer comprising memory means and being connected to said at least one temperature sensor and said at least one pressure sensor, said memory means storing a predefined thermodynamic model and a predefined mathematical model, the predefined thermodynamic model providing a calculation of the compressor isentropic efficiency from measured compressor temperature and pressure values, the predefined mathematical model being empirically calibrated using historical temperature and pressure data relating to the compressor in a full-load and stable mode of the latter, said at least one temperature sensor being configured to measure at least the compressor inlet temperature, said at least one pressure sensor being configured to measure at least the compressor inlet pressure, the method comprising an initial step of measuring the compressor inlet temperature and the compressor inlet pressure, the method comprising the computer-implemented following steps:

receiving a first measured value corresponding to the compressor inlet temperature, and a second measured value corresponding to the compressor inlet pressure;
calculating the compressor isentropic efficiency from the first and second measured values, using the predefined thermodynamic model;
predicting the compressor isentropic efficiency from the first and second measured values, using the predefined mathematical model;
calculating the real time difference between the calculated compressor isentropic efficiency and the predicted compressor isentropic efficiency;
based on the calculated real time difference, providing at least one decision indicator as to whether or not a compressor offline washing is necessary.

The method according to the invention may comprise one or more of the following features considered alone or in any combination that is technically possible:

said at least one temperature sensor is configured to further measure the compressor outlet temperature, said at least one pressure sensor is configured to further measure the compressor outlet pressure, the method further comprises a step of receiving a third measured value corresponding to the compressor outlet temperature and a fourth measured value corresponding to the compressor outlet pressure and, during the calculating step, the compressor isentropic efficiency is calculated from the first, second, third and fourth measured values, using the predefined thermodynamic model;
said predefined mathematical model comprises a single-layer non-convolutional neural network performing a linear regression;
said linear regression comprises a third-degree polynomial transform of the data;
the method further comprises a step of filtering a series of calculated real time differences for predefined nominal and stable periods of the compressor;
the method further comprises a step of averaging said results of the filtering step, and wherein the step of providing at least one decision indicator comprises a first sub-step of comparing said calculated average or a time derivative of a series of calculated averages to a predefined threshold, and a second sub-step during which, if said calculated average or said time derivative of a series of calculated averages is greater to said predefined threshold, the decision indicator is set to indicate an offline washing of the compressor;
the method further comprises a step of calculating a difference between the maximum value and the minimum value among said results of the filtering step, and wherein the step of providing at least one decision indicator comprises a third sub-step of comparing said calculated difference to a second predefined threshold, said second predefined threshold being lower to the first predefined threshold, and a fourth sub-step during which, if the calculated average or the time derivative of a series of calculated averages is lower to the first predefined threshold and if said calculated difference is lower to said second predefined threshold, the decision indicator is set to indicate no offline washing of the compressor;
the step of providing at least one decision indicator comprises a fifth sub-step during which, if the calculated average or the time derivative of a series of calculated averages is lower to the first predefined threshold and if said calculated difference is greater to said second predefined threshold, variability of the input data is compared to a predefined variability threshold, and if said variability is lower to said predefined variability threshold, the decision indicator is set to indicate an offline washing of the compressor.

The invention also relates to a computer program product comprising a readable information carrier having stored thereon a computer program comprising program instructions, the computer program being loadable onto a data processing unit and causing at least the steps of receiving a first and a second measured values, calculating the compressor isentropic efficiency, predicting the compressor isentropic efficiency, calculating the real time difference between the calculated compressor isentropic efficiency and the predicted compressor isentropic efficiency and providing at least one decision indicator of a method as previously described to be carried out when the computer program is carried out on the data processing unit.
The invention also relates to a readable information carrier on which is stored a computer program product as previously described.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be easier to understand in view of the following description, provided solely as an example and with reference to the appended drawings in which:

Figure 1 is a schematic view of an example of an assembly for implementing a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, said assembly including a computer, and
Figure 2 is a flowchart of an example of implementation of a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly.

DETAILED DESCRIPTION OF SOME EMBODIMENTS

An example of an assembly 18 comprising a calculator 20, a computer program product 22 stocked thereon, a first temperature sensor 23A, a second temperature sensor 23B, a first pressure sensor 23C and a second pressure sensor 23D are illustrated on figure 1. The first temperature sensor 23A is configured to measure the compressor inlet temperature, and the second temperature sensor 23B is configured to measure the compressor inlet pressure. The first pressure sensor 23C is configured to measure the compressor outlet temperature, and the second pressure sensor 23D is configured to measure the compressor outlet pressure.
The calculator 20 is connected to the four sensors 23A-23D and is preferably a computer.
More generally, the calculator 20 is a computer or computing system, or similar electronic computing device adapted to manipulate and/or transform data represented as physical, such as electronic, quantities within the computing system's registers and/or memories into other data similarly represented as physical quantities within the computing system's memories, registers or other such information storage, transmission or display devices.
The calculator 20 interacts with the computer program product 22.
As illustrated on figure 1, the calculator 20 comprises a processor 24 comprising a data processing unit 26, memories 28 and a reader 30 for information media. In the example illustrated on figure 1, the calculator 20 comprises a human machine interface 32, such as a keyboard, and a display 34. The memories 28 store a predefined thermodynamic model 35 and a predefined mathematical model 37.
The calculator 20 further comprises an information medium 36, on which is stored the computer program product 22.
The information medium 36 is a medium readable by the calculator 20, usually by the data processing unit 26. The readable information medium 36 is a medium suitable for storing electronic instructions and capable of being coupled to a computer system bus.
By way of example, the information medium 36 is a USB key, a floppy disk or flexible disk (of the English name "Floppy disc"), an optical disk, a CD-ROM, a magneto-optical disk, a ROM memory, a memory RAM, EPROM memory, EEPROM memory, magnetic card or optical card.
On the information medium 36 is stored the computer program 22 comprising program instructions.
The computer program 22 is loadable on the data processing unit 26 and is adapted to entail the implementation of a method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, when the computer program 22 is loaded on the processing unit 26 of the calculator 20.
A method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, will now be described with reference to figure 2. In the description, "offline washing of a compressor" is defined as the washing of the mechanical parts of the compressor when said compressor is not in operation and when the turbine-compressor assembly is offline. Neither the turbine-compressor assembly nor the compressor are shown in the figures, for clarity purposes.
The predefined thermodynamic model 35 provides a calculation of the compressor isentropic efficiency from measured compressor temperature and pressure values. As a non-limiting example, when the compressor inlet temperature, the compressor outlet temperature, the compressor inlet pressure and the compressor outlet pressure values are available, the thermodynamic model 35 is parameterized to provide the calculation of the compressor isentropic efficiency from the following equation (1): $C_{e} = \frac{(\Pr^{\frac{γ - 1}{γ}} - 1)}{Tr - 1} * 100$ where Ce is the compressor isentropic efficiency, Pr is the pressure ratio between the compressor outlet pressure and the compressor inlet pressure, Tr is the temperature ratio between the compressor outlet temperature and the compressor inlet temperature and γ is a numerical factor substantially equal to 1,4.
The predefined mathematical model 37 is empirically calibrated using historical temperature and pressure data relating to the compressor in a full-load and stable mode of the latter. More precisely, the predefined mathematical model 37 is a physics-informed machine learning model, or more specifically a physics-informed neural network.
Physics-Informed Machine Learning integrates knowledge from data and mathematical physics models. Machine Learning models can learn from governing physical rules and hence provide physically consistent priors that can be integrated as a bias or constraints in a data-based learning. If we consider a specific task, these physics priors can be embedded in the design choices made on the Neural Network architecture. For example, some physics principles like symmetry, invariance to rotations, shifts and other geometrical transformations can be reflected by some Neural Networks and their kernels like Convolution Neural Networks, Covariant Neural Networks or equivariant transformer networks. Some other physics properties can be also encoded in network the architecture (neuron structures, activations, etc.).
Hence, the predefined mathematical model 37 is a model already trained, calibrated and validated for its long-term physics consistency predictions.
Preferably, the predefined mathematical model 37 is first trained on the basis of historical data in a full-load and stable mode of the compressor and then fine-tuned on the basis of more realistic data like reanalysis data.
Preferably, the predefined mathematical model 37 comprises a single-layer non-convolutional neural network performing a linear regression. More preferably, the linear regression comprises a third-degree polynomial transform of the data. As a non-limiting example, the mathematical model 37 is parameterized to provide a prediction of the compressor isentropic efficiency from the following equation (2): E_ce = 0.0012913174630341416 + 0.0060008691115246073 * Ti + 0.4207440902711638 * Pi + 0.30185402681107676 * Ti ² - 0.008520932313727378 * Ti* Pi- 0.0005005281368429298 * Pi ² + 8.892233927335477e ^-05 * Ti ³ - 0.0003159434683341549 * Ti ² * Pi + 8.962081050559727e^-06 * Ti * Pi ² + 1.6778738147263305e ^-07 * Pi ³ where Ece is the prediction of the compressor isentropic efficiency, Pi is the compressor inlet pressure and Ti is the compressor inlet temperature.
The optimizing method comprises an initial step 110 which consists of the four sensors 23A-23D respectively measuring the compressor inlet temperature, the compressor inlet pressure, the compressor outlet temperature and the compressor outlet pressure.
The optimizing method then comprises a step 120 which consists of the calculator 20 receiving a first measured value corresponding to the compressor inlet temperature, a second measured value corresponding to the compressor inlet pressure, a third measured value corresponding to the compressor outlet temperature and a fourth measured value corresponding to the compressor outlet pressure. The first measured value is provided by the first temperature sensor 23A, the second measured value is provided by the second temperature sensor 23B, the third measured value is provided by the first pressure sensor 23C and the fourth measured value is provided by the second pressure sensor 23D.
The optimizing method then comprises a step 121 which consists of the processor 24 of the calculator 20 calculating the compressor isentropic efficiency from the first, second, third and fourth measured values, using the predefined thermodynamic model 35. More precisely, in a particular embodiment, the processor 24 of the calculator 20 computes the compressor isentropic efficiency using the above-mentioned equation (1).
The optimizing method comprises a step 122 which consists of the processor 24 of the calculator 20 predicting the compressor isentropic efficiency from the first and second measured values, using the predefined mathematical model 37. More precisely, in a particular embodiment, the processor 24 of the calculator 20 predicts the compressor isentropic efficiency using the above-mentioned equation (2).
It is to be noted that the calculating step 120 and the predicting step 122 can be performed in parallel, or alternatively the calculating step 120 can be performed before the predicting step 122, or alternatively the predicting step 122 can be performed before the calculating step 120.
The optimizing method then comprises a step 124 which consists of the processor 24 of the calculator 20 calculating the real time difference between the compressor isentropic efficiency calculated at step 120 and the compressor isentropic efficiency predicted at step 122. It is to be noted that the calculation at step 124 is performed on a single event (only one point per event).
Optionally, the optimizing method comprises a step 126 which consists of the calculator 20 filtering a series of real time differences calculated at step 124 for predefined nominal and stable periods of the compressor (to this end, the calculator 20 includes digital filtering means, not shown in the figures, for clarity purposes). In the description, "nominal and stable periods of the compressor" is defined as the periods during which the compressor operates in a stable and nominal mode. As a non-limiting example, it can be considered empirically (based on historical data) that the compressor operates in a stable and nominal mode after substantially 30 minutes of the first part of the compressor cycle.
The optimizing method then comprises a step 128 which consists of the processor 24 of the calculator 20 averaging the results output from the filtering step 126.
The optimizing method then comprises a step 130 which consists of the processor 24 of the calculator 20 comparing the average calculated at step 128 or a time derivative of a series of calculated averages calculated at step 128 to a first predefined threshold.
If said calculated average or said time derivative of a series of calculated averages is greater to the first predefined threshold, the optimizing method then comprises a step 132A during which the processor 24 of the calculator 20 sets a decision indicator to indicate an offline washing of the compressor. For example, the processor 24 of the calculator 20 sets this decision indicator onto the display 34. If not, at a step 134, the processor 24 of the calculator 20 sets the decision indicator to indicate no offline washing of the compressor.
If said calculated average or said time derivative of a series of calculated averages is lower to the first predefined threshold, the optimizing method then comprises a step 134 during which the processor 24 of the calculator 20 calculates a difference between the maximum value and the minimum value among the results output from the filtering step 126.
The optimizing method then comprises a step 136 which consists of the processor 24 of the calculator 20 comparing the difference calculated at step 134 to a second predefined threshold. The second predefined threshold is lower to the first predefined threshold.
If said calculated difference is lower to the second predefined threshold, the optimizing method then comprises a step 138A during which the processor 24 of the calculator 20 sets the decision indicator to indicate no offline washing of the compressor.
If said calculated difference is greater to the second predefined threshold, the optimizing method then comprises a step 140 which consists of the processor 24 of the calculator 20 comparing variability of the input data to a predefined variability threshold. If said variability of the input data is lower to the predefined variability threshold, the optimizing method then comprises a step 142A during which the processor 24 of the calculator 20 sets a decision indicator to indicate an offline washing of the compressor. For example, the processor 24 of the calculator 20 sets this decision indicator onto the display 34. If not, at a step 142B, the processor 24 of the calculator 20 sets the decision indicator to indicate no offline washing of the compressor.
The method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly according to the invention is cost effective, energy saving, and prevents too much degradation of the compressor by avoiding doing too many washings or not enough ones.
The person skilled in the art will understand that the embodiments and variants described above can be combined to form new embodiments provided that they are technically compatible.

Claims

A method for optimizing a procedure for triggering an offline washing of a compressor in a turbine-compressor assembly, the method being implemented by an assembly including a computer (20), at least one temperature sensor (23A, 23B) and at least one pressure sensor (23C, 23D), said computer (20) comprising memory means (28) and being connected to said at least one temperature sensor (23A, 23B) and said at least one pressure sensor (23C, 23D), said memory means (28) storing a predefined thermodynamic model (35) and a predefined mathematical model (37), the predefined thermodynamic model (35) providing a calculation of the compressor isentropic efficiency from measured compressor temperature and pressure values, the predefined mathematical model (37) being empirically calibrated using historical temperature and pressure data relating to the compressor in a full-load and stable mode of the latter, said at least one temperature sensor (23A) being configured to measure at least the compressor inlet temperature, said at least one pressure sensor (23C) being configured to measure at least the compressor inlet pressure, the method comprising an initial step of measuring (110) the compressor inlet temperature and the compressor inlet pressure, the method comprising the computer-implemented following steps:
- receiving (120) a first measured value corresponding to the compressor inlet temperature, and a second measured value corresponding to the compressor inlet pressure;

- calculating (121) the compressor isentropic efficiency from the first and second measured values, using the predefined thermodynamic model (35);

- predicting (122) the compressor isentropic efficiency from the first and second measured values, using the predefined mathematical mode (37);

- calculating (124) the real time difference between the calculated compressor isentropic efficiency and the predicted compressor isentropic efficiency;

- based on the calculated real time difference, providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator as to whether or not a compressor offline washing is necessary.
A method according to claim 1, wherein said at least one temperature sensor (23A, 23B) is configured to further measure the compressor outlet temperature, said at least one pressure sensor (23C, 23D) is configured to further measure the compressor outlet pressure, the method further comprises a step (120) of receiving a third measured value corresponding to the compressor outlet temperature and a fourth measured value corresponding to the compressor outlet pressure and, during the calculating step (121), the compressor isentropic efficiency is calculated from the first, second, third and fourth measured values, using the predefined thermodynamic model (35).
A method according to claim 1 or 2, wherein said predefined mathematical model (37) comprises a single-layer non-convolutional neural network performing a linear regression.
A method according to claim 3, wherein said linear regression comprises a third-degree polynomial transform of the data.
A method according to any one of claims 1 to 4, wherein the method further comprises a step of filtering (126) a series of calculated real time differences for predefined nominal and stable periods of the compressor.
A method according to claim 5, wherein the method further comprises a step of averaging (128) said results of the filtering step (126), and wherein the step of providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator comprises a first sub-step of comparing (130) said calculated average or a time derivative of a series of calculated averages to a predefined threshold, and a second sub-step (132A) during which, if said calculated average or said time derivative of a series of calculated averages is greater to said predefined threshold, the decision indicator is set to indicate an offline washing of the compressor.
A method according to claim 6, wherein the method further comprises a step of calculating (134) a difference between the maximum value and the minimum value among said results of the filtering step (126), and wherein the step of providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator comprises a third sub-step of comparing (136) said calculated difference to a second predefined threshold, said second predefined threshold being lower to the first predefined threshold, and a fourth sub-step (138A) during which, if the calculated average or the time derivative of a series of calculated averages is lower to the first predefined threshold and if said calculated difference is lower to said second predefined threshold, the decision indicator is set to indicate no offline washing of the compressor.
A method according to claim 7, wherein the step of providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator comprises a fifth sub-step (140) during which, if the calculated average or the time derivative of a series of calculated averages is lower to the first predefined threshold and if said calculated difference is greater to said second predefined threshold, variability of the input data is compared to a predefined variability threshold, and if said variability is lower to said predefined variability threshold, the decision indicator is set (142A) to indicate an offline washing of the compressor.
A computer program product (22) comprising a readable information carrier having stored thereon a computer program comprising program instructions, the computer program (22) being loadable onto a data processing unit (26) and causing at least the steps of receiving (120) a first and a second measured values, calculating (121) the compressor isentropic efficiency, predicting (122) the compressor isentropic efficiency, calculating (124) the real time difference between the calculated compressor isentropic efficiency and the predicted compressor isentropic efficiency and providing (128, 130, 132A, 134, 136, 138A, 140, 142A, 142B) at least one decision indicator of a method according to any one of claims 1 to 8 to be carried out when the computer program (22) is carried out on the data processing unit (26).
A readable information carrier on which a computer program product (22) according to claim 9 is stored.