Patent Application
20240289522
The present disclosure relates to a method and system for fatigue analysis and, more particularly, a method and system for analyzing the fatigue life and damage state of a polymeric physical asset.
Solutions for fatigue analysis from Finite Element Analysis (FEA) of components have been available for many years. Nonlimiting examples of fatigue analysis solutions for metallic components are described in each of:
For many, FEA has become an essential part of maturing and qualifying design concepts. However, once these designs are finalized, it can be difficult to adequately track and predict the life cycle of a product due to actual loads and damage accrued, and when maintenance may be required. As used herein, the term “predict” is associated with the post-production context of a product, as opposed to being used in the pre-production context. Materials, mechanical parts, and structural supports generally require periodic inspection to ensure safety and function.
In traditional reactive maintenance methods, field engineers fix problems after faults are detected. Today's common practice is a “paper and pencil” based fixed schedule maintenance, which mainly relies on the experience of engineers responsible for the maintenance, and on visual inspection of the physical asset. These engineers often must review run time sensor data from the devices manually.
In theory, it is important to inspect the historical trending data to estimate the wear and tear of a specific machine. However, in practice, the historical data may show no problems exist before a catastrophe occurs, or the engineer may not be able to access the data on a vendor's server. Additionally, the physical asset is often not accessible for inspection, and damage may be difficult to detect on the physical asset. Thus, engineers are operating on limited information, resulting in inefficiencies and imprecision in maintenance.
There is a continuing need for a method and system for efficiently obtaining load histories at potential failure locations by incorporating FEA calculations into a simulation. Desirably, the system and method combine run time sensor data with the FEA-based system to more accurately predict when and where maintenance is required.
In concordance with the instant disclosure, a method and system for efficiently obtaining load histories at potential failure locations by incorporating FEA calculations into a digital twin simulation, and which combines run time sensor data with the FEA-based digital twin system to more accurately predict when maintenance is required, has been surprisingly discovered.
Assets tend to accumulate damage due to operating loads, exposure to the environment, and the like. Owners and operators often do not have accurate or complete information on remaining life, current state of damage, acceptable maintenance intervals, prior warning of failure. This leads to unexpected loss, failure, downtime, safety issues. Operating loads can be determined and recorded by an appropriate combination of sensors and computational modeling. The present disclosure includes a system for using at least one of actual and virtual load history to incrementally update a damage model of the asset.
Advantageously, and to create a more efficient process relative to the prior art, run time sensor data can be used according to the present method and system to create or update a “digital twin.” The term “digital twin” refers to a digital replica of physical assets, processes or systems that can be used to predict the life cycle and maintenance requirements of an item. The present system for using digital twins for predictive maintenance allows an engineer to know if an article requires maintenance before any wear or damage is noticed. The digital twin model may be implemented to execute any number of simulations, resulting in a model that accurately predicts degradation, end of life, and damage events.
The present method and system may periodically receive operational information from an asset, (e.g., recorded by load, displacement, temperature, acceleration sensors) and compute, for each period, the updated, current state of damage occurring in the asset. The system maintains current the damage state of a digital twin to reflect the effects of all operating history received from the sensors. The residual life may also be computed after each period, in terms of repeats of a hypothetical ideal load case, or in terms of repeats of the total history experienced previously. The system also may automatically generate status reports, maintenance reminders, diagnostics and warnings based upon computed estimates of remaining fatigue life. Importantly, fatigue calculations are also based on critical plane analysis.
In one embodiment, a digital twin system for predicting a residual life of a physical asset includes at least one user computer and at least one server. The at least one user computer has a graphical user interface permitting a user to receive at least one of a damage event warning, an end of life warning, and a status report. The at least one server in communication with the at least one user computer. The at least one server includes at least one processor and at least one memory. The at least one memory includes a tangible, non-transitory computer readable medium with processor-executable instructions stored thereon. The at least one server includes an administration subsystem in communication with a data source. The administration subsystem has at least one database and configured to receive operating history data of the physical asset from the data source and store the operating history data of the physical into the at least one database. The at least one server also includes a simulator in communication with the administration subsystem. The at least one server is configured for storing the digital twin of the physical asset. The digital twin includes a model of the physical asset and a current damage state. The at least one server is also configured for receiving a periodic residual life simulation request from the administration subsystem, receiving the operating history data of the physical asset from the administration subsystem, and updating the digital twin of the physical asset with the operating history data of the physical asset, using a fatigue solver algorithm to update the current damage state of the digital twin, thereby providing an updated digital twin. A residual life prediction for the physical asset is generated by using the fatigue solver algorithm with a hypothetical operating history data and the updated digital twin. The residual life prediction is indicative of the residual life before reaching a failure mode associated with the physical asset in operation.
The data source of the digital twin system may include at least one physical sensor in communication with the physical asset. The at least one sensor may be configured to monitor at least one of load, displacement, temperature, and acceleration of the physical asset. The physical asset may be formed from a polymeric or elastomeric material. The administration subsystem of the present disclosure may continuously and automatically receive the operational history data from the at least one sensor. The periodic residual life simulation requests themselves may occur at least one of once per minute, once per hour, and once per day. The data source may also be manual user input as the operating history data into the system via the at least one user computer.
The model may particularly be a finite element analysis model. The simulator may further comprise an interpolation engine. The fatigue solver algorithm may further include a critical plane analysis. For example, the fatigue solver algorithm may be defined by:
wherein Δc is a change in crack length, i is a time period, j is an element of the model, k is a plane orientation, r is a crack growth rate, T is an energy release rate, εmn is a strain tensor history, θ is a temperature history, c is a crack length, and N is cycles.
The hypothetical operating history data may also include cycles of a hypothetical ideal load case. The hypothetical operating history data may also be cycles of a total operating history of the physical asset.
In certain instances, the simulator automatically generates at least one of the damage event warning, the end of life warning, and the status report to the at least one user computer where a predetermined condition occurs. For example, the predetermined condition may include at least one of where the size of a crack exceeds a maximum size and where remaining cycles of an ideal hypothetical load case exceeds a minimum threshold. The user may further input the predetermined condition by manually inputting the predetermined condition into the system via the at least one user computer.
In additional instances, the data source may include a structural dynamics simulation. The structural dynamics simulation may use operational history of a second physical asset that is in communication with the physical asset to generate the operational history of the physical asset.
In another embodiment, a method for predicting a residual life of a physical asset includes the steps comprising providing the digital twin system and storing, by the simulator, the digital twin of the physical asset the at least one memory. The simulator receives the periodic residual life simulation from the administration subsystem. The simulator also receives the operating history data of the physical asset from the administration subsystem. The simulator further updates the digital twin of the physical asset with the operating history data of the physical asset, using a fatigue solver algorithm to update the current damage state of the digital twin, thereby providing an updated digital twin. The simulator then generates a residual life prediction for the physical asset by using the fatigue solver algorithm with a hypothetical operating history data and the updated digital twin. The residual life prediction is indicative of the residual life before reaching a failure mode associated with the physical asset in operation.
In an exemplary embodiment, a digital twin system for predicting a residual life of a physical asset includes at least one user computer and at least one server. The user computer has a graphical user interface permitting a user to receive damage event warnings, end of life warnings, and status reports.
The server communicates with the user computer and includes at least one memory. The memory includes a tangible, non-transitory computer readable medium with processor-executable instructions stored thereon. Additionally, the memory includes an administration subsystem that is in communication with a data source. The administration subsystem includes at least one database. The administration subsystem is configured to receive the physical asset's operating history data transmitted by the data source and store the operating history data into the database.
The memory further includes a simulator. The simulator is in communication with the administration subsystem and is configured to perform several functions, e.g. receiving periodic residual life simulation requests and operating history data of the physical asset from the administration subsystem. A further function of the simulator is storing the digital twin of the physical asset. This digital twin includes a model of the physical asset with the current damage state of the digital twin.
Another function of the simulator is to update the digital twin and generate residual life predictions. The simulator updates the digital twin by updating the current damage state of the digital twin using a fatigue solver algorithm and the operating history data of the physical asset. The simulator generates a residual life prediction for the physical asset by using the fatigue solver algorithm with a hypothetical operating history data and the updated digital twin.
In yet another embodiment, a method for predicting the residual life of a physical asset includes the first step of providing the digital twin system of the embodiment disclosed above. A second step of the simulator storing the digital twin of the physical asset. The digital twin includes the model of the physical asset with the current damage state of the physical asset. A third step of the simulator receiving the periodic residual life simulation request from the administration subsystem. A fourth step of the simulator receiving the operating history data of the physical asset from the administration subsystem.
A fifth step of the simulator updating the digital twin by updating the current damage state of the digital by using the fatigue solver algorithm with the operating history data of the physical asset. A sixth step of the simulator generating a residual life prediction for the physical asset by using the fatigue solver algorithm with a hypothetical operating history data and the updated digital twin. The residual life prediction is indicative of the residual life before reaching a failure mode associated with the physical asset in operation.
In a particular embodiment, the system disclosed above is modified to separate out the administration subsystem into its own administration server and the simulator into its own simulator server. The exemplary embodiment further includes at least one physical sensor as the data source. The physical sensor is in communication with the physical asset and is configured to monitor the load, displacement, temperature, and acceleration of the physical asset. Additionally, a user could act as the data source by manually inputting the operating history data into the system using the user computer.
The hypothetical operating history data is cycles of a hypothetical ideal load case or cycles of a total operation history of the physical asset. Also, the simulator server is further configured to generate the damage event warnings, end of life warnings, and status reports to the user computer where the size of a crack exceeds a maximum size or where remaining cycles of ideal hypothetical load case exceeds a minimum threshold. Additionally, the simulator server further comprises a critical plane analysis with the fatigue solver algorithm. The fatigue solver algorithm used by the simulator server is:
The equation has the following variables: Δc is the change in crack length, i is the period, j is the element of the model, k is the plane orientation, r is the crack growth rate, Tis the energy release rate, εmn is the strain tensor history, θ is the temperature history, c is crack length, and N is the cycles.
Further areas of applicability will become apparent from the description provided herein. It should be understood that the description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.
The above, as well as other advantages of the present disclosure, will become readily apparent to those skilled in the art from the following detailed description, particularly when considered in the light of the drawings described herein.
The following detailed description and appended drawings describe and illustrate various embodiments of the invention. The description and drawings serve to enable one skilled in the art to make and use the invention and are not intended to limit the scope of the invention in any manner. In respect of the methods disclosed, the order of the steps presented is exemplary in nature, and thus, is not necessary or critical unless otherwise disclosed.
A digital twin system 2 for predicting a residual life of a physical asset 6 is shown in
It should be appreciated that the at least one server has sufficient processing power and memory in order to store and process the run time data, FEA, and digital twin models as described herein in a timely manner. For example, on a simulator server 14 with 24G of RAM and six (6) processing cores, processing thirty (30) minutes of operating history data sampled at 500 Hz (roughly 1 million time steps) for 50,000 elements used 6 MB of RAM per finite element for the fatigue analysis. Although these minimum hardware requirements have been found especially useful for timely processing according to the present method, other suitable hardware requirements for processing the method of the present disclosure may also be selected by the skilled artisan, as desired.
The administration server 4 is in communication with the at least one sensor 8, the asset manager 12, the user terminal 10, and the simulator server 14. The administration server 4 includes a tangible, non-transitory computer readable medium with processor-executable instructions stored thereon. The memory includes at least one database. One function of the administration server 4 is acting as a database that stores an initial damage state of the physical asset 6 and an operating history data of the physical asset 6.
It should be understood that the physical asset 6 of the present disclosure is any asset comprising, consisting of, or consisting essentially of polymeric or elastomeric materials. As non-limiting examples, this may include bushings, seals, or tires. Other types of polymeric or elastomeric products may also be employed as the physical asset 6, as desired.
As used herein, the refers to “initial damage state” a predetermined state associated with each finite element of a model of the physical asset 6. The initial damage state may be provided in the form of a user-defined file that contains material definitions, output requests and strain history associated with the physical asset 6. The operating history data may include variables that are measurements of time, load, displacement, and temperature over a specific or predetermined time period. One skilled in the art may include more variables to monitor additional aspects of the physical asset 6.
Another function of the administration server 4 is to automatically transmit status reports, maintenance reminders, diagnostics and warnings to the asset manager 12 based upon the residual life predictions generated by the simulator server 14 where the predictions exceed a predetermined boundary. For example, where the size of a crack exceeds a maximum size and where remaining cycles of an ideal hypothetical load case exceeds a minimum threshold, the administration server 4 may be configured to automatically transmit a warning signal.
The at least one sensor 8 is in communication with the physical asset 6 and the administration server 4. The at least one sensor 8 monitors the physical asset 6 and collects the operating history data. For example, the at least one sensor 8 may include a physical sensor disposed on or adjacent to or in communication with the physical asset 6, or otherwise configured to monitor, at least one of time, load, displacement, and temperature of the physical asset 6 in operation. This data collection by the at least one sensor 8 may be performed automatically and may be continuously or intermittently transmitted to the administration server 4 to permit the processes for updating the digital twin to occur. Optionally, the operational history data can be manually inputted into the digital twin system 2 by a user via the user terminal 10. This is useful when a sensor 8 malfunctions or a sensor 8 isn't available to measure a specific variable.
The simulator server 14 is in communication with the administration server 4 and includes a tangible, non-transitory computer readable medium with processor-executable instructions stored thereon. One function of the simulator server 14 is periodically receive residual simulation requests and the operating history data from the administration server 4. The administration server 4 may periodically generate these requests, for example, once per minute, once per hour, once per day, or any other periodicity selected by a skilled artisan. Further functions of the simulation server 14 may include, as non-limiting examples: storing the digital twin of the physical asset 6; creating or updating the digital twin to synchronize its damage state with the physical asset 6; and generating residual life predictions for the physical asset 6.
In operation, as shown in
The finite element models of the present disclosure include a set of individual finite elements, each of which contains information about stresses and strains, and also the size of cracks in each element, for example, as shown in
One suitable method and system for creating a finite element model was described in U.S. Pat. Appl. Publication No. 2004/0254772 to Su, the entire disclosure of which is hereby incorporated herein by reference. Other suitable methods for acquiring and processing load data or other time-varying load data may also be used within the scope of the present disclosure.
With renewed reference to
In preferred embodiments, the fatigue solver algorithm 22 is based upon the principles of critical plane analysis. Critical plane analysis refers to the analysis of stresses or strains as they are experienced by a particular plane in a material, as well as the identification of which plane is likely to experience the most extreme damage. Critical plane analysis may be used in engineering to account for the effects of cyclic, multiaxial load histories on the fatigue life of materials and structures. Different critical plane analyses are described in the following references, the entire disclosures of which are also hereby incorporated herein by reference.
The fatigue solver algorithm 22 has the following variables:
The simulator server 14 writes the result of the fatigue solver algorithm 22 to a file called a “restart file,” identified by reference number “44” in
The fifth column, STIFFNESS 56, is a multiplier of an original stiffness to account for cyclic softening in operation. The sixth column, EMBRITTLEMENT 58, is a multiplier of the original stiffness to account for aging over the operational history. The seventh column, CRACKSIZE_MIN 60, is a smallest crack size associated with the critical plane obtained during the prior analysis (i.e., the immediately prior restart file 44). The eighth column, CRACKSIZE_AVG 62, is the average crack size from the critical plane search of the prior analysis. The ninth column, CRACKSIZE_MAX 64, is the largest crack size from the critical plane search of the prior analysis.
Referring now to
As shown in
It should be appreciated that the incremental procedure described hereinabove readily accounts for effects associated with the order in which events occur with respect to the physical asset 6 in operation. For example, a series of severe loads occurring early in service of the physical asset 6 may induce greater (or lesser) damage than the same loads applied later in life, depending on, for example, the Mullins effect and mode of control details, among other variables.
After the simulator server 14 creates or updates the digital twin as shown in
Next, in a step 204, the simulator server 14 incorporates the hypothetical operating history data 66 into a model such as a finite element model using FEA 18. Then, in a step 206, the simulator server 14 incorporates the updated damage state 46 of the physical asset 6 and the finite element model parameters into the fatigue solver algorithm 22. A hypothetical updated damage state is then output from the fatigue solver algorithm 22 in a step 208, which is then used to repeat the same process until the hypothetical updated damage state is within the boundaries of a predefined or predetermined failure mode for the physical asset 6. Likewise, the hypothetical operating history data may be supplied and used in each of steps 210-224 shown in
A practical application of the residual life prediction 68 methodology is shown in
For example, vehicle “A” could hypothetically endure 1.39E6 repeats of the routine case immediately following the installation load history, and that number of repeats is reduced following the application of each new operation in the history. The abuse events applied to vehicle “A” are seen to produce an especially large drop in life left. Vehicle A at the end of all operations only has 4.40E4 repeats of the routine case remain. Although vehicle “B” has undergone a large number of routine events, it has a larger remaining life at the end of the schedule because it did not experience the abuse case.
In another embodiment, the digital twin system can further comprise an interpolation engine 70 as shown in
In particular,
In a further embodiment, the digital twin system can further comprise a method 400 involving a structural dynamics simulation 72, for example, as shown in
It should be understood that the system 2 and methods 100, 200, 300, 400 of the present disclosure efficiently obtain load histories at potential failure locations of physical assets and incorporate FEA calculations into digital twin simulations. The system 2 and methods 100, 200, 300, 400 may further combine run time sensor data with the FEA-based digital twin system, as detailed hereinabove, to more accurately predict when maintenance is required for physical assets 6 having the digital twins.
While certain representative embodiments and details have been shown for purposes of illustrating the invention, it will be apparent to those skilled in the art that various changes may be made without departing from the scope of the disclosure, which is further described in the following appended claims.
This application claims the benefit of U.S. Provisional Application No. 62/567,354, filed on Oct. 3, 2017. The entire disclosure of the above application is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 62567354 | Oct 2017 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 16149738 | Oct 2018 | US |
| Child | 18594882 | US |
