Technical field
[0001] The present invention relates to a method for fatigue mapping a physical asset. More specifically it relates to a method for fatigue mapping a physical asset using a digital twin of the physical asset.
Background
[0002] It is often desirable to assess and/or predict operational parameters regarding the properties of a mechanical system to predict safe and economical maintenance.
[0003] Different technological areas use several different maintenance schemes for predicting and performing maintenance. Some of the common schemes is either preventive maintenance (PM), predictive maintenance, reliability-centered maintenance (RCM) or condition-based maintenance (CBM) or a combination of thereof. Typically, maintenance intervals or exchange intervals for a component or a system may be set as a given time by the manufacture. Said methods does not take into consideration what type of use the component or system is exposed to, for instance if the usage has been more extreme than normal, fatigue failure will occur earlier.
Therefore, systems not taking into consideration the magnitude of a maximum force or number of cycles the components or a system sees during a time period, must include huge safety margins in the maintenance schedule.
[0004] Or, the maintenance schedule can be decided on by reactive methods, where problems are solved, and parts and assets are changed when faults are detected. This type of run to failure maintenance is not suitable in critical system where an error in a component would be fatal.
[0005] For the conventional methods used for determining the maintenance of mechanical systems and machinery, decisions are based on limited information and on-the-safe-side mentality, resulting increased cost, additional impact on the environment and reduced efficiency.
[0006] Other maintenance assessment methods include the use of computer models of an asset, such as a digital twin, where simulations are run on the digital twin to better predict optimal maintenance. US2016247129 discloses a system for using digital twins for scalable, model-based machine predictive maintenance, wherein digital twins corresponding to remotely located physical machines, a database comprising run time log data collected from sensors associated with the physical machine, a simulation platform configured to process simulation models corresponding to the digital twins using a computer system. However, the system does not take into the consideration the placement for the sensor nor does the system calculate the measurement error in the sensor or the where on the physical asset the critical areas may be. Furthermore, the system does not indicate critical areas or areas of interests on the physical asset before sensor data is collected. The disclosed system in US2016247129 does not indicate any real time and/or instant monitoring or real time and/or instant maintenance assessment. Document US2019/0102494 A1 discloses a method for fatigue mapping a physical object that involves feeding sensor data from a physical object to a numeric model and performing calculations on this numerical model based on the sensor data. Document EP 3348983 A1 discloses a method for monitoring the position of a physical asset. The method teaches a system for using multiple virtual sensors on a digital twin in addition to physical sensors to limit the number of physical sensors [0003] to calculate where a physical object is located. From [0010] it is learned that the virtual sensors can be placed where physical sensors do not physically fit (for example inside a solid steel beam) and that many virtual sensors can be used without increasing the costs associated with hardware and sensors. EP 346226 A1 discloses a method for controlling a physical asset. The documents teach how to determine a system response associated with the failure modes of the electro-mechanical system from the sensor data, and that the system model may be validated based on the real-time sensor data. None of the disclosed known systems or methods provides any means for calculations before the sensors are located nor do they disclose both a detailed real time fatigue mapping and an analysis of the future fatigue condition of a physical asset.
[0007]It is therefore an aim of the present invention to provide a method and system for maintenance assessment where the drawbacks mentioned is avoided.
Brief Description of The Disclosure
[0008]The invention relates to a method and system for a fatigue mapping process of physical assets. The method comprises the steps of; establishing a digital twin of the asset, wherein the digital twin comprises a numerical model of the asset and expose the digital twin to a modeled load scenario and/or modeled forces. Further, the method comprises the steps of calculate, by a processor, critical areas of the digital twin based on a set of criterions to identify high stress areas and/or components of the digital twin, wherein the criterions for identifying high stress areas and/or components in step b) includes at least one of the following criterions; locating local maximum and/or local minimum and/or global maximum and/or global minimum values of stress or strain, and outfit the physical asset with at least one sensor on at least one of the identified high stress areas. Data is register from said at least one sensor, and transfer to the digital twin. Real-time calculations are performed, by a processor, on the digital twin, wherein the sensor data from the physical asset is input to the calculations, wherein the calculations comprising fatigue calculations on the digital twin based on the real time feed-back data and determining the condition of the physical asset based on at least one condition parameter. The system Comprises a computer device comprising memory storage and a processor. Wherein the processor is configured to perform a set of calculations on a numerical model based on load data and output the results from said calculations. The system further comprises a graphical interface engine adapted to display a rendering of a digital twin and/or the results from numerical calculations, at least one sensor, signal transmitting means for the at least one sensor, signal receiving means for the computer device and at least one physical asset.
Brief Description of the Drawings
Fig 1. shows an example of a physical asset in the form of a wind turbine.
Fig. 2 displays a graphical rendering of a digital twin of a wind turbine.
Fig. 3 shows a close up of the digital twin graphically rendered.
Fig. 4 shows a representation of the system that may be provided in accordance with some embodiments of the invention.
Fig. 5 shows a block diagram of a fatigue mapping method in accordance with some embodiments of the invention.
Fig. 6 shows a block diagram of the system that may be provided in accordance with some embodiments of the invention.
Description of fatigue mapping
[0009]Fatigue mapping is a method for predicting the remaining fatigue life (RFL) of a structure, also known as the remaining usefully life (RUL), subjected to typical dynamic loading. Dynamic loads are load conditions that will vary over time. These dynamic loads are hard to predict because both the magnitude and the variation in time for the loads are affected by a great number of unpredictable variables. For instance, when evaluating the RFL of a suspension spring for a motor vehicle, is hard to predict if the spring has seen easy road usage i.e. lower dynamic loads, or hard off-road use i.e. higher dynamic loads, and what the future usage of the suspension spring might be subject to.
[0010] Where previous concepts for maintenance where based on history, experience, time or track-records. Said invention is a system end method for mapping the predicted remaining life based on the dynamic load history of the physical asset and the simulation of the system linked to big data management. Linking the simulation of the system to big data is a way of big data management, where large quantities’ quantitative or qualitative data is utilized as management and decision tools. Fatigue mapping in accordance with the invention will lead to a new design philosophy where future concepts and future physical assets may be designed and engineered with lower safety factor because of the greater knowledge of the actual subjected loads. This will lead to reduction in material use and more environmentally friendly designs.
[0011] The data is gathered from sensors, as sensors are collecting time histories of the usage of the asset, and information is given for both time and magnitude. In this way, the used fatigue life of the asset will be monitored. This is based on 1. Sensed data, 2. digital twin model, 3. fatigue material data relative to that particular alloy or material prone to fatigue failure. A system will then analyze the monitored data and link that to the digital twin.
[0012] The dynamic load history of a physical asset is gathered through at least one sensor located on said physical asset, wherein the at least one sensor gathers information in real time. One of the advantages of the herein disclosed invention of fatigue mapping is that it is not strictly necessary to place the at least one sensor(s) at the exact critical area(s). As the advantages described can be achieved when there is a correlation between the sensor data and a calculated computer result from a digital twin. Thereby, the critical locations, i.e. the locations on a physical asset with higher stress loads or higher probability of failure, will be predicted and calculated on a digital twin regardless of the placements of the sensors.
[0013] Material fatigue may not lead to any visible or noticeable wear on a physical asset or component and may never be noticed or measured before what might be considered a fatal or devastating accident has happened. A digital twin would be able to, based on at least one correctly allocated sensor on a physical unit, count the number of cycles or number of impact loads that said unit would have been subjected to, and from said sensor information, calculate and map the material fatigue situation of the unit The calculations would only increase in accuracy as the digital twin gathers data over time. The at least sensor on each unit will count and register how many cycles and how high, or severer, material fatigue the unit have been subject to, and the digital twin will, based on that, measure how many cycles remains before a potential fatal and devastating error occurs.
[0014]Examples of structures subjected to dynamic loads is rotating structures such as ship- or aircraft shafts and propeller, helicopters and helicopter parts, gear boxes, gas turbines, aircraft jet engines, hydropower turbines or wind turbine blades.
Furthermore, complex mechanical systems like engines including suspension brackets, gearbox components and dynamic pressure vessels may be subject to dynamic loads from multiple directions simultaneously. It should be obvious that several other systems and structures are subjected to dynamic loadings, and that the method disclosed herein could be used on all such systems.
Description of Digital twin
[0015]When a physical asset is delivered from a producer or factory to its user or customer, an exact digital twin may be delivered as well. The digital twin is identical in every possible detail and may in some cases be specific for the exact physical asset, for instance can manufactory anomalies and actual measured physical and geometrical properties of the physical asset be implemented to the digital twin. The digital twin will be a hierarchical, computational structures model of the physical asset. The digital twin will accept probabilistic input of all load types, static- and dynamical loads, environmental, and usage factors, and it also tightly couples to an as-built computational fluid dynamics (CFD) model. In some embodiments of the inventions the digital twin is a computational fluid dynamic (CFD) numerical analysis model or model by means of the finite element method (FEM). The format of the FEM -model depends on the program used to generate the model. This may be, but not limited to, known formats such as ANSYS, NASTRAN, ABAQUS or SESAM. In some cases, neutral formats such as STEP, IGES, Parasolid or ACIS may be most beneficial.
However, it would be obvious for a person skilled in the art that other computer models, formats, data structures and numerical calculation methods may be used.
[0016] The digital twin is used for virtual simulations of real-world scenarios where simulation data is gathered. The data comprises material performance, structural performance, damage data and fatigue data. In addition to a fatigue model, the digital twin may be comprised of several different models such as a thermal model and a dynamic load model for increasing the accuracy of linked and codependent parameters.
[0017] The digital twin may initially, in the design fatigue life mapping process, be used to map critical values, wherein the critical values comprises numerical values and locations for critical areas of the physical asset. The critical areas are found by analyzing failure criteria’s and critical components and situations. When the critical values are calculated, the locations for physical monitoring is decided on based on said map of critical values.
[0018] By first performing an initial mapping of critical values of the digital twin of the physical asset, the sensor measurements measured by a sensor on the physical asset at a later stage, can be used to validate the digital twin and sensor placement. By first obtaining calculations from a model and then obtaining real time measurements from at least one senor, the results can be compared. Said comparison of simulated values and measured values is further used to either validate the computer calculations and models of the digital twin, or to indicate that changes needs to be made to computer calculations and models of the digital twin. To map the correlation between each physical asset and its digital twin is paramount to achieve the advantages over known prior art. After a correlation has been verified, continuous feed-back in the form of sensor measurements and monitoring data will be linked to the digital twin to predict the status of RFL.
[0019] The advantages of the invention disclosed herein is that of reducing the number of unplanned maintenance stops, reducing the number of unnecessary maintenance stops, reducing unnecessary loads on the environment. Further advantages are to prolong the operational life of a physical asset or physical system and reducing maintenance costs. Select the most optimal materials for the physical asset or physical system. Additional further advantages are to optimize systems design and optimize design philosophy.
Detailed description of the invention
[0020] Below, various embodiments of the invention will be described with reference to the figures, in which like numerals in different figures describes the same features.
[0021]Fig 1. shows a physical asset 1, in this example a wind turbine comprising a turbine part 7 and rotors 8a, 8b, 8c. The rotors 8a, 8b, 8c are in this case stiffened with strut bars 9a, 9b, 9c. The physical asset is equipped with sensors 4, in this example the sensors are strain gauges (SG1-SG8).
[0022] In fig. 2 a digital twin 2 displayed graphically as a representation of a numerical model of the wind turbine in fig.1. The sensor 4 shown in the figure as strain gauge SG3 is located at a local maximum of measured strain, graphically represented by a different color gradient that the surrounding areas. Said graphical representation of change color gradient is a representation of a critical areas or areas of interest.
[0023]Fig. 3 shows a cut out of the digital twin 2, where a limited portion of the rotor blade 8a and the attachment of the strut bar 9a is graphically represented by a different color gradient. The area marked by the strain gauge SG4 is displayed with a different gradient that the surrounding areas. During the initial phase of the fatigue mapping process the digital twin numerical model is first used to calculate the critical areas, or areas of interest expected, or modeled, loads. This initial simulation will give the result for where critical areas might be and where to place the sensors 4. A benefit of this initial simulation is that the result can be used to verify that the results from the simulations on the numerical model corresponds to the values measured on the physical asset, and hence to verify the validity the digital twin 2 of the physical asset 1.
[0024] Fig. 4 shows a representation of the system comprising a physical asset 1 represented by a wind turbine comprising an electric generator 7, rotors 8a, 8b, 8c. The digital twin 2 is in this case displayed graphically on a graphical interface 5, such as a typical digital screen. The graphical interface 5 in fig. 4 displays a portion of digital representation of the digital twin 2. The system shown in fig. 4 comprises a computer device 3 comprising memory storage 3a and a processor 3b. The computer device is a typically computer adapted for simulations and numerical calculations and rendering of the result from said simulations and calculations. Which further components said computer is equipped with or is comprised of, would be obvious for a person skilled in the art. The computer 3 is configured to perform a set of calculations on the numerical model based on anticipated, expected or measured load data, and output the results from said calculations to a graphical interface 5 or graphical interface engine that displays a rendering of representation the digital twin 2 and/or displays the results from calculations performed by the computer.
[0025] Fig. 5 shows a block diagram of a fatigue mapping method in accordance with some embodiments of the invention. The method comprises the steps a-g, wherein the first step a) establishing a digital twin in a form previously described by asserting or compiling a computer model, wherein the digital twin comprises a numerical model of the asset, and b) calculate preferred sensor positions with digital twin by exposing the digital twin to a modeled load scenario and/or modeled forces and calculate, by a processor, critical areas of the digital twin based on a set of criterions to identify high stress areas and/or components of the digital twin, and c) outfit the physical asset with sensors on at least one of the identified high stress areas from the previous step.
Additionally, the method comprises step d) to registering data from said at least one sensor, and e) transfer the sensor data to the digital twin by feeding the sensor data through communicational means from the sensor to the processing device and/or communication to the digital twin, and f) perform real-time calculations, by a processor, on the digital twin, wherein the sensor data from the asset is input to the calculations, wherein the calculations comprising fatigue calculations on the digital twin based on the real time feed-back data, and step g) for determining the condition of the physical asset based the results from the fatigue calculations.
[0026]Fig. 6 shows a block diagram illustrating the cyclic process of fatigue mapping as a dynamic system in accordance with an embodiment of the invention. The process may start with a method in accordance with the one described in fig. 5 and a new cycle will start at step b) except the load data fed to the digital twin is the results from the fatigue calculations from step g) not a modeled load. With is cyclical method, the calculations for the modeled fatigue mapping will increase in accuracy as fatigue mapping process continues. Each cycle will increase the accuracy of the calculations and the duration of the predicament. A new cycle of the process will start and the digital twin calculation model, now updated with the most resent fatigue analysis of the remaining useful life, will be fed with new real time data from at least one sensor, new calculations will be reformed on the digital twin calculation model and new results from the calculations will yield new results for the remaining useful life-
[0027]Further in Fig. 4, the signals from the sensor 4 is transferred from the physical asset 1 to the computer device 3 via transmitting means 6. In the embodiment illustrated in Fig. 4 the signal transmitting means 6 are wireless transmitting means 6 comprising a wireless signal transmitter 6a connected to or in communication with the at least one sensor 4, the sensor 4 being placed on the physical asset 1, to transmit signals from the at least one sensor 4 wirelessly to a wireless signal receiver 6b connected to or in communication with the computer device 3.
[0028] To perform the fatigue mapping method a physical asset 1, first it has to be established a digital twin 2 of the physical asset 1. The digital twin 2 comprises a numerical model of the physical asset. As previously stated, the numerical model may be a computational fluid dynamic (CFD) numerical analysis of a finite element method, where the results of said analysis may be rendered graphically by a graphical interface 5. The computation comprises exposing the digital twin 2 to a modeled load scenario and/or modeled forces and perform calculations by a computer device 3 comprising memory storage 3a and a processor 3b. By exposing the digital twin 2 to modeled loads or model scenarios should be understood as feeding input parameters corresponding to real values to said calculation, as to reflect the actual values of the real world. The result of said calculations can then be used to identify critical areas of the digital twin 2. The critical areas identified on the digital twin 2 will thus correspond to the same areas on the physical asset. For example, critical areas or areas of interest, may be identified by identifying local maximum or minimum values of stress or strain, or global local maximum or minimum values of stress or strain. And these local or global areas will represent an area of interest or a critical area. These local or global maximums or minimums is thus used as criterions to identify high stress areas and/or components of the digital twin. Once these areas of interest or critical areas has been identified on the digital twin 2, the physical asset is outfitted with at least one sensor 4 or multiple sensors 4 on at least one of the identified areas. It can either be multiple sensors of one type or different types. Typically, the sensor(s) 4 may be a strain gauge to measure strain i.e. tension or compression of an area, a temperature gauge to measure temperature and/or an accelerometer to measure the acceleration, or rate of change of velocity at a certain point. To measure vibrations on the physical asset 1 either an accelerometer or proximity probes may be used.
Proximity probes are noncontacting transducers that measure distance to a target and the change and/or rate of change, in distance to a that target. Proximity probes is therefore used to measure for example vibration on objects that cannot have sensors fixed to them, such on moving objects ant the likes of rotating machinery and drive shafts of engines. This at least one sensor(s) 4 or multiple different sensors 4 produces sensor data when installed. This registered data from said at least one sensor 4 is transferred from the at least one sensor to the digital twin 2.
[0029] With a numerical model and data from the sensors attached to the physical asset 1, the step of performing calculations by the use of a computer 3 device comprising memory storage 3a and a processor 3b, on the digital twin, wherein the sensor data from the asset is input to the calculations, wherein the calculations comprising fatigue calculations on the digital twin based on the feed-back senor data. Further input to the calculations and the numerical model may be probabilistic input of all load types, static- and dynamical loads, environmental, and usage factors. This data may be real time data gathered from the at least one sensor, historical data from the least one sensor gathered over time or a combination. The results from said calculations will then determine the condition of the physical asset based on the condition parameters. There maty be multiple different parameters for the condition to determine, or map, the fatigue situation of a physical asset 1. These may be remaining usefully life or remaining fatigue life of a component or the asset, number of cycles to failure, magnitude of static load before failure, duration and magnitude of dynamic load before failure
[0030] Although specific embodiments of the invention have been described and illustrated herein, it is recognized that modifications and variations may readily occur to those skilled in the art, and consequently, it is intended that the claims be interpreted to cover such modifications and equivalents.
Reference numerals
1 Physical asset, wind turbine
2 Digital twin
3 Computer device
3a Memory storage
3b Processor
4 Sensor(s) illustrated as SG1-SG8 5 Graphical interface, display screen 6 Signal transmitting means
6a Wireless signal transmitter
6b Wireless signal receiver
7 Electric generator of wind turbine 8a-8c Rotor blades of wind turbine
9a-9c Strut bars of wind turbine