A method for constructing a reduced-order model of fluid-structure interaction based on digital twins

Abstract

The present invention relates to a method for constructing a fluid-solid coupling reduced-order model based on digital twins, which includes: establishing physical models of fluid domain and solid domain, processing the mesh common nodes and performing mesh division, suppressing the solid domain model, simulating and solving the full-order model of the fluid domain mesh, using the inlet velocity as the excitation and the pressure change of the fluid-solid coupling surface as the output response to construct the fluid reduced-order model; suppressing the fluid domain model, simulating and solving the full-order model of the solid domain mesh, also using pressure as the excitation and stress and displacement as the output response to construct the solid reduced-order model; using the pressure change response output by the fluid reduced-order model as the input excitation of the solid reduced-order model, a fluid-solid coupling reduced-order model combining the fluid reduced-order model with the solid reduced-order model can be obtained. Compared with the traditional fluid-solid coupling method, the present invention can shorten the calculation time of the fluid-solid coupling reduced-order model from hours to seconds.

Description

Technical field:

[0001] The present invention relates to the field of digital technology and marine engineering, and specifically to a method for constructing a fluid-solid coupling reduced-order model based on digital twins. Background technology:

[0002] With the vigorous development of society, the contradiction between human beings' desire to improve the quality of life and the shortage of land resources is becoming increasingly severe. Countries around the world have gradually shifted the focus of resource acquisition to the ocean, which accounts for 71% of the earth's surface area. More and more countries have taken the rational and orderly development and utilization of marine energy and the protection of the marine environment as the basic national policy for pursuing development. With the acceleration of the development of marine energy, marine equipment such as offshore platforms have also been built on a large scale. Among them, the exploration and development of marine resources, especially the jacket-type offshore platforms for the development of oil and natural gas, account for the majority. So far, about 2,000 large and medium-sized jacket-type offshore platforms have been built in the world, and their working water depth has reached four or five hundred meters. Due to the complex and changeable marine environment, the pile legs of the jacket-type offshore platform have been eroded by environmental loads for a long time during service. The disasters and accidents caused by this may cause property loss, casualties, natural pollution, and even affect international reputation. Therefore, in order to ensure the safety of the jacket-type offshore platform in the harsh environment of the deep sea, it is necessary to monitor its pile leg structure in real time and predict its life, which involves the fluid-solid coupling problem between the pile leg structure and the ocean, which puts extremely high requirements on the simulation calculation speed.

[0003] A large number of studies on fluid-solid coupling of jacket-type offshore platforms have been conducted by predecessors. For example, in 2023, Science, Technology and Engineering published "Dynamic Response Analysis of Jacket Platforms under Extreme Storm Loads". This paper conducted a dynamic analysis of jacket-type offshore platforms under extreme pulsating wind conditions and discussed the stress, strain and other characteristics of the platform under different extreme storm conditions. In 2015, Ship Science and Technology published "The Influence of Wave-Current Coupling Force on the Wedge Block of the Offshore Platform Leg". This paper established the leg structure of the local offshore platform using the finite element method, and applied environmental loads such as wave force and wind force to study the influence of wave-current coupling force on the strength of the leg structure. In 2005, Journal of Ocean University of China published "Nonlinear Dynamic Analysis of Offshore Platform Structure Considering Fluid-Solid Coupling". This paper proposed a nonlinear dynamic equation of an offshore platform considering fluid-solid coupling and its time-history analysis method, and studied the difference in dynamic response of the platform with or without fluid-solid coupling. Although there have been major breakthroughs in the research on fluid-solid coupling of jacket-type offshore platforms, its inefficient simulation speed has limited the development of this field. The concept of digital twins has scientifically solved the time-consuming computational shortcomings of traditional simulation methods. The Chinese invention patent application with application number 202310354023.1 discloses a detection method based on a digital twin reduced-order model of a transformer. According to the operating status of the transformer to be inspected, the Internet of Things is used to construct its digital twin reduced-order model to obtain the transient temperature field. The Chinese invention patent application with application number 20161064029.4 discloses a free vibration analysis method for a beam structure with a reduced-order model. The basis vectors are constructed by polynomial interpolation functions to achieve reduced-order processing of the original structure. Although many achievements have been made in model reduction, the current research on the application of digital twin reduced-order model methods in fluid-solid coupling of jacket-type offshore platforms at home and abroad is still relatively insufficient.

[0004] In summary, fluid-solid coupling analysis of the leg structure of a jacket-type offshore platform is of great significance to the safe operation of the platform. However, the existing technology usually performs calculations in hours or even days when solving the numerical simulation operation problems of the fluid-solid coupling of the leg structure of the platform. The calculation time is long and consumes a lot of resources, and it is impossible to quickly respond to the deformation of the leg structure of the platform. Due to the lack of methods that can be used to construct a reduced-order model for the rapid calculation of the fluid-solid coupling of the leg structure of the platform, this method is not very effective in engineering applications. Therefore, how to accurately and timely respond to the fluid-solid coupling problem of the leg structure of a jacket-type offshore platform has become a difficult problem. Summary of the invention:

[0005] The purpose of the present invention is to provide a method for constructing a fluid-solid coupling reduced-order model based on digital twins. This method for constructing a fluid-solid coupling reduced-order model based on digital twins is used to solve the problem that the prior art has a long calculation time and consumes a lot of resources when solving the numerical simulation calculation problem of fluid-solid coupling of platform pile leg structures, and is unable to quickly respond to the deformation of the platform pile legs.

[0006] The technical solution adopted by the present invention to solve the technical problem is: the method for constructing a fluid-solid coupling reduced-order model based on digital twins comprises the following steps:

[0007] Step 1: Establish the physical model of fluid domain and solid domain, perform mesh common node processing and mesh division, suppress the solid domain model, and output the fluid domain mesh;

[0008] Step 2: Suppress the fluid domain of the fluid domain and solid domain physical model, and release the suppression of the solid domain to obtain a full-order model of the solid domain mesh;

[0009] Step 3: Simulate and solve the full-order model of the fluid domain mesh obtained in step 1. Take the inlet velocity as the excitation and the pressure change on the fluid-solid coupling surface as the output response. Carry out numerical simulation to obtain the excitation vector C(t) and response vector B(t) for model order reduction. The expression is as follows:

[0010] C(t)={V1(t);V2(t);…;V i (t)} (1)

[0011] B(t)={n1(t);n2(t);…;n i (t)} (2)

[0012] Where V i (t) is the excitation vector component; n i (t) is the response vector component;

[0013] Step 4: Perform dynamic order reduction on the excitation vector C(t) and response vector B(t) obtained in step 3 to determine a dynamic system, which can be expressed as follows:

[0014] B(t)=f(B(t),C(t)) (3)

[0015] B| t＝0 =B0 (4)

[0016] Where B(t) is the derivative of the response vector B(t); B0 is the initial condition, i.e., the solution of the first time step; f is a nonlinear function of the response vector B(t) and the excitation vector C(t), which is an iterative process with the goal of minimization until it reaches the set accuracy a:

[0017] min||B ROM (t)-B(t)||2≤a (5)

[0018] In the formula, B ROM (t) is the reduced-order response vector; a is the accuracy value set according to the actual needs of the user;

[0019] Determine the main modal orders that characterize the full-order model of the fluid domain and reconstruct it to obtain a reduced-order model of the fluid;

[0020] Step 5: Simulate and solve the full-order model of the solid domain mesh obtained in step 2, take the pressure change applied to the fluid-solid coupling surface as the excitation, take the stress and displacement of the solid domain as the output response, carry out numerical simulation, and perform order reduction processing to obtain a solid reduced-order model;

[0021] Step 6: Use the pressure change response output by the fluid reduced-order model obtained in step 4 as the input excitation of the solid reduced-order model obtained in step 5 to obtain a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model.

[0022] Beneficial effects:

[0023] 1. The calculation time of the fluid-solid coupling reduced-order model described in the present invention can be shortened from hours to seconds, with low calculation cost. It can quickly respond to the stress and displacement of the offshore platform pile legs, overcome the drawbacks of traditional fluid-solid coupling numerical simulation calculations, take into account the computational efficiency and calculation accuracy of numerical simulation, have good engineering applicability, and provide a digital twin numerical simulation solution for traditional fluid-solid coupling simulation calculations.

[0024] 2. The present invention takes the local jacket-type offshore platform pile leg structure under the action of fluid pressure in the fluid as the physical model carrier of numerical simulation, and proposes a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model. The fluid-solid coupling reduced-order model uses velocity as excitation and solid structure stress and displacement as response. The calculation time can be shortened to seconds, and rapid response can be made according to different velocity excitations. Description of the drawings:

[0025] Figure 1 It is a local jacket-type offshore platform pile leg model under the action of fluid pressure in the fluid;

[0026] Figure 2 Name the plots for the fluid and solid domain walls;

[0027] Figure 3 Comparison diagram of fluid-solid coupling surface pressure between the fluid reduced-order model and the full-order model;

[0028] Figure 4 This is a displacement comparison diagram of the solid reduced-order model and the full-order model;

[0029] Figure 5 This is the framework structure diagram of the fluid-solid coupling reduced-order model;

[0030] Figure 6 It is a schematic diagram of the process of the present invention. Specific implementation method:

[0031] The present invention will be further described below in conjunction with the accompanying drawings:

[0032] Combination Figure 6 As shown in the figure, this method for constructing a fluid-solid coupling reduced-order model based on digital twins: establish physical models of the fluid domain and the solid domain, perform mesh common node processing and mesh division on them, suppress the solid domain model, simulate and solve the full-order model of the fluid domain mesh, and import the simulation results into the ANSYS Twin Builder software, use the inlet velocity as the excitation, and the pressure change on the fluid-solid coupling surface as the output response to construct a fluid reduced-order model; suppress the fluid domain model, simulate and solve the full-order model of the solid domain mesh, and also import the simulation results into the ANSYS Twin Builder software, use pressure as the excitation, and use stress and displacement as the output responses to construct a solid reduced-order model; use the pressure change response output by the fluid reduced-order model as the input excitation of the solid reduced-order model, and you can get a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model.

[0033] Specifically:

[0034] Step 1: Preprocess the physical model in ANSYS Workbench. In order to ensure the common nodes of the meshes of the fluid domain and the solid domain, set the fluid domain and the solid domain as the same component file, then mesh the physical model, suppress the solid domain of the processed physical model, retain only the fluid domain mesh, and output the fluid domain mesh.

[0035] Step 2: Suppress the fluid domain of the physical model and release the suppression of the solid domain.

[0036] Step 3: Use Fluent software to simulate and solve the fluid domain grid obtained in step 1. The inlet uses a velocity inlet that changes sinusoidally with time, and the outlet uses a free outlet. The inlet velocity is used as the excitation, and the pressure change of the fluid-solid coupling surface wall_fsi is used as the output response, and then the csv file and bin file for model reduction are obtained, which can be defined as the excitation vector C(t) and the response vector B(t), expressed as follows:

[0037] C(t)={V1(t);V2(t);…;V i (t)} (1)

[0038] B(t)={n1(t);n2(t);…;n i (t)} (2)

[0039] Where V i (t) is the excitation vector component; n i (t) is the response vector component.

[0040] Step 4: Based on ANSYS Twin Builder software, the csv file and bin file obtained in step 3 are dynamically reduced in order. According to the accuracy selection principle, the main modal orders that can characterize the full-order model of the fluid domain can be determined, and these main modes are reconstructed to obtain the fluid reduced-order model, as follows:

[0041] The excitation vector C(t) and response vector B(t) obtained in step 3 are dynamically reduced to form a dynamic system, which can be expressed as follows:

[0042] B(t)=f(B(t),C(t)) (3)

[0043] B| t＝0 =B0 (4)

[0044] Where B(t) is the derivative of the response vector B(t); B0 is the initial condition, i.e., the solution of the first time step; f is a nonlinear function of the response vector B(t) and the excitation vector C(t). This function is an iterative process with the goal of minimization until it reaches the set accuracy a. The process can be expressed as follows:

[0045] min||B ROM (t)-B(t)||2≤a (5)

[0046] In the formula, B ROM (t) is the reduced-order response vector; a is the accuracy value set according to the actual needs of the user.

[0047] According to the selection principle of formula (5), the main modal orders that can characterize the full-order model of the fluid domain can be determined, and these main modes can be reconstructed to obtain the fluid reduced-order model.

[0048] Step 5: Perform numerical calculations on the solid domain model obtained in step 2, set the surface wall_bottom as a fixed surface, and apply a pressure that changes sinusoidally with time to the surface wall_fsi. Use the pressure change applied to the surface wall_fsi as the excitation, and the stress and displacement of the solid domain as the output response. Perform numerical simulations and obtain csv files and bin files for model reduction.

[0049] Step 6. Also based on ANSYS Twin Builder software, the csv file and bin file obtained in step 5 are processed to reduce the order of the output response. According to the accuracy selection principle in step 4, the main modal order that can characterize the full-order model of the solid domain can be determined, and these main modes can be reconstructed to obtain a solid reduced-order model.

[0050] Step 7: Use the pressure change response output by the fluid reduced-order model obtained in step 4 as the input excitation of the solid reduced-order model obtained in step 6, so as to obtain a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model.

[0051] Example:

[0052] Step 1: First, construct the physical model by CAD to obtain Figure 1 The local jacket-type offshore platform pile leg model under the action of fluid pressure in the fluid shown in the figure, the naming method of each wall of the physical model is as follows Figure 2 As shown. The physical model is preprocessed in ANSYS Workbench, the fluid domain is set to fluid format, and the solid domain is set to solid format. In order to ensure the common nodes of the mesh of the fluid domain and the solid domain, the fluid domain and the solid domain are set to the same component file, and then the physical model is meshed as a whole, and the meshing solver is set to CFD format. The solid domain of the processed physical model is suppressed, only the fluid domain mesh is retained, and the fluid domain mesh is output;

[0053] Step 2: Suppress the fluid domain of the physical model and release the suppression of the solid domain to construct a subsequent solid reduced-order model;

[0054] Step 3. Perform numerical simulation on the fluid domain mesh obtained in step 1 through Fluent software, and compile (extension-load "rbp") command to activate the ROM Builder Pre function. The Realizable k-ε model is used to simulate the turbulent flow of the fluid. The inlet uses a velocity inlet that changes sinusoidally with time, and the outlet uses a free outlet. Take the inlet velocity as the excitation, and the pressure change on the fluid-solid coupling surface wall_fsi as the output response. Carry out numerical simulation and obtain the csv file and bin file for fluid domain model reduction. The csv file contains information related to the input excitation, and the bin file contains information related to the output response, which can be defined as the excitation vector C(t) and the response vector B(t), expressed as follows:

[0055] C(t)={V1(t);V2(t);…;V i (t)} (1)

[0056] B(t)={n1(t);n2(t);…;n i (t)} (2)

[0057] Where V i (t) is the excitation vector component; n i (t) is the response vector component;

[0058] Step 4: Based on ANSYS Twin Builder software, the excitation vector C(t) and response vector B(t) obtained in step 3 are dynamically reduced to form a dynamic system, which can be expressed as follows:

[0059] B(t)=f(B(t),C(t)) (3)

[0060] B| t＝0 =B0 (4)

[0061] Where B(t) is the derivative of the response vector B(t); B0 is the initial condition, i.e., the solution of the first time step; f is a nonlinear function of the response vector B(t) and the excitation vector C(t). This function is an iterative process with the goal of minimization until it reaches the set accuracy a. The process can be expressed as follows:

[0062] min||B ROM (t)-B(t)||2≤a (5)

[0063] In the formula, B ROM (t) is the reduced-order response vector; a is the accuracy value set according to the actual needs of the user.

[0064] First, the output response is reduced in order, and the accuracy value a is set to 2%. According to the selection principle of formula (5), under the constraint of accuracy, only the first sixteen modes of the response are retained to characterize the full-order model of the fluid domain, and then the fluid reduced-order model can be obtained. The calculation time using the traditional CFD method is about 8 hours, and the calculation time using the reduced-order model method can be shortened to 1 second. Comparing the pressure results of the fluid-solid coupling surface between the fluid reduced-order model and the full-order model at each time point, as shown in Table 1, it can be verified that the fluid reduced-order model has high accuracy. At 1 second, the pressure comparison of the fluid-solid coupling surface between the fluid reduced-order model and the full-order model is Figure 3 As shown;

[0065] Table 1:

[0066]

[0067]

[0068] Step 5: Perform numerical calculation on the solid domain model of step 2, set the bottom surface of the pile leg wall_bottom as a fixed surface, and apply a pressure that changes sinusoidally with time to the surface wall_fsi. Activate the DynaROM Pre function, use the pressure change applied to the surface wall_fsi as the excitation, and use the stress and displacement of the solid domain as the output response. Carry out numerical simulation and obtain the csv file and bin file for the solid domain model order reduction, where the csv file contains information related to the input excitation, and the bin file contains information related to the output response;

[0069] Step 6. Also based on ANSYS Twin Builder software, use the dynamic order reduction method to process the csv file and bin file obtained in step 5, reduce the order of the output response, set the accuracy value a to 1%, and according to the selection principle of formula (5) in step 4, under the constraint of accuracy, only retain the first-order mode of the response to characterize the full-order model of the solid domain, and then obtain the solid reduced-order model. The calculation time using the traditional CFD method is about 3 hours, and the calculation time using the reduced-order model method can be shortened to 1 second. Compare the stress results of the solid reduced-order model and the full-order model at each time point, as shown in Table 2; compare the displacement results of the solid reduced-order model and the full-order model at each time point, as shown in Table 3, which verifies that the solid reduced-order model has high accuracy. At 1 second, the displacement comparison between the solid reduced-order model and the full-order model is Figure 4 shown.

[0070] Table 2:

[0071]

[0072]

[0073] Table 3:

[0074]

[0075]

[0076] Step 7: Use the pressure change response output by the fluid reduced-order model obtained in step 4 as the input excitation of the solid reduced-order model obtained in step 6 to obtain a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model. The framework structure is as follows: Figure 5 As shown in the figure, the fluid-solid coupling reduced-order model uses velocity as the excitation and solid structure stress and displacement as the response. Compared with the traditional fluid-solid coupling method, the numerical simulation calculation time of the fluid-solid coupling reduced-order model can be shortened to seconds, and it can respond quickly according to different stimuli.

Claims

1. A method for constructing a fluid-structure interaction reduced-order model based on digital twins, characterized in that The following steps are involved: Step 1: Establish the physical model of fluid domain and solid domain, perform mesh common node processing and mesh division, suppress the solid domain model, and output the fluid domain mesh; Step 2: Suppress the fluid domain of the fluid domain and solid domain physical model, and release the suppression of the solid domain to obtain a full-order model of the solid domain mesh; Step 3: Simulate and solve the full-order model of the fluid domain mesh obtained in step 1. Take the inlet velocity as the excitation and the pressure change on the fluid-solid coupling surface as the output response. Carry out numerical simulation to obtain the excitation vector C(t) and response vector B(t) for model order reduction. The expression is as follows: C(t)={V1(t);V2(t);...;V i (t)} (1) B(t)={n1(t);n2(t);...;n i (t)} (2) Where V i (t) is the excitation vector component; n i (t) is the response vector component; Step 4: Perform dynamic order reduction on the excitation vector C(t) and response vector B(t) obtained in step 3 to determine a dynamic system, which can be expressed as follows: B′(t)=f(B(t),C(t)) (3) B| t＝0 =B0 (4) Where B′(t) is the derivative of the response vector B(t); B0 is the initial condition, i.e., the solution of the first time step; f is a nonlinear function of the response vector B(t) and the excitation vector C(t). This function is an iterative process with the goal of minimization until it reaches the set accuracy a: min||B ROM (t)-B(t)||2≤a (5) In the formula, B ROM (t) is the reduced-order response vector; a is the accuracy value set according to the actual needs of the user; Determine the main modal orders that characterize the full-order model of the fluid domain and reconstruct it to obtain a reduced-order model of the fluid; Step 5: Simulate and solve the full-order model of the solid domain mesh obtained in step 2, take the pressure change applied to the fluid-solid coupling surface as the excitation, take the stress and displacement of the solid domain as the output response, carry out numerical simulation, and perform order reduction processing to obtain a solid reduced-order model; Step 6: Use the pressure change response output by the fluid reduced-order model obtained in step 4 as the input excitation of the solid reduced-order model obtained in step 5 to obtain a fluid-solid coupling reduced-order model that combines the fluid reduced-order model with the solid reduced-order model.

Images