Corrosive environment FRP-concrete interface mechanism model correction method

Abstract

The invention discloses a corrosion environment FRP-concrete interface mechanism model correction method which comprises the following steps: S1, extracting original data of related literatures, determining a typical interface corrosion type, and establishing a test database of multiple influence factors; s2, establishing an FRP-concrete interface finite element model, and modeling an interface shear test piece; and S3, performing inversion correction on the interface constitutive model, determining and optimizing a Latin hypercube sampling method through comparison and selection, performing iterative calculation, and finally obtaining an inversion analysis result. According to the method, through the whole inversion correction process, the optimized Latin hypercube sampling method and the gradient optimization method are used for model correction, the advantages and disadvantages of the two model correction methods are compared, fine correction and rapid correction methods of the interface constitutive model are determined, the influence of strain test errors on test results is avoided, and the test accuracy is improved. And a bond-slip constitutive relation capable of accurately reflecting the mechanical properties of the corrosion interface is established.

Description

Technical Field

[0001] The present invention relates to the technical field of research on interface degradation mechanism, in particular to a method for modifying the interface mechanism model of FRP-concrete in a corrosive environment. Background Art

[0002] Urban renewal and rural dilapidated house renovation are the key development directions of the current construction industry. With the development of society, people's requirements for the safety of building service are constantly increasing. Due to environmental corrosion and long-term use, the structure of old buildings deteriorates, making it unable to meet the strength and stiffness requirements. Therefore, safe and reliable technologies are needed to reinforce and strengthen the structure.

[0003] Fiber Reinforced Polymer (FRP), as an excellent engineering reinforcement material, has the advantages of light weight, high strength, good corrosion resistance, etc., and is widely used in structural reinforcement. The improvement of the bearing capacity of FRP-reinforced structures depends on the effectiveness of interfacial stress transfer. Therefore, the bond performance of the FRP reinforcement interface has become the focus of reinforcement research. In actual engineering, FRP-reinforced concrete structures often bear the action of corrosive environments, such as freeze-thaw, wet-heat, acid mist, and salt mist. The degree of deterioration of the interfacial bond performance is much higher than that of the FRP and concrete base materials. The interfacial bond layer may become the weak layer of the entire reinforcement interface, and premature interfacial peeling failure of the FRP may occur, resulting in a sharp decline in the bearing capacity and stiffness of the reinforced structure, seriously affecting the safety of the reinforced structure.

[0004] There are great controversies in the existing research on the interface constitutive models under various corrosion actions. Through years of experimental research on the FRP-concrete interface by researchers, it is found that due to factors such as the non-homogeneity of concrete and discontinuous strain testing, there are serious differences in the interface constitutive models obtained by testing at different positions of the interface. Therefore, there are problems with the effectiveness of the interface constitutive models obtained solely by experimental testing.

[0005] Carrying out research on the durability of the bond performance of the FRP-concrete interface based on experiments and finite element inverse correction, establishing an interface test library for corrosive environments, and combining model correction technology to establish a constitutive model that can truly reflect the durability degradation performance of the interface is of great significance for effectively understanding the degradation mechanism of the interface bond performance under corrosive conditions, carrying out durability design of the FRP-concrete reinforcement interface, and ensuring the safety of the reinforced structure.

[0006] Reinforced structures are often exposed to corrosive working environments, such as salt corrosion, wet-dry cycles, freeze-thaw cycles, and ultraviolet radiation. With the extension of the exposure time, the bond performance of the interface deteriorates, seriously affecting the safety performance of the reinforced structure. Therefore, the durability problem of the FRP-concrete interface under corrosive environments is a research hotspot in the current reinforcement field.

[0007] Most of the current experimental studies on the constitutive relationship of the FRP-concrete interface have inevitable errors. The sources of errors mainly include: the inhomogeneity of the concrete matrix surface, and the discontinuous testing of strain gauges, etc. The inhomogeneity of the concrete matrix refers to the uneven distribution of strains at each measuring point caused by the random distribution of aggregates and pores on the concrete surface; in addition, due to the discontinuous testing of strain gauges, only the average stress within the gauge length can be collected, and the maximum interfacial shear stress cannot be guaranteed to be collected within the gauge length. Because of the errors in the strain test results, the errors in the interfacial bonding stress obtained by differentiation are relatively large.

[0008] Due to the limitations of the experimental testing method, there are certain problems with the interfacial bond-slip constitutive relationship obtained through strain testing. When we substitute these constitutive models into the interface of the finite element, we often find that there are serious deviations between the load-slip curve and the experiment, indicating that the constitutive models obtained from these experiments cannot accurately reflect the bonding performance of the interface.

[0009] In most of the existing studies on the performance of the FRP-concrete interface in a corrosive environment, when substituting the interfacial constitutive relationship obtained from the experiment into the finite element model, there are serious discrepancies between the load-slip curves of the model and the experiment. To solve this problem, it is necessary to study the rapid correction method of the constitutive model. Based on the corrected constitutive model, study the degradation mechanism of the interfacial bonding performance under corrosion.

[0010] Aiming at the problem that the existing interfacial bonding performance model under corrosion cannot accurately invert the interfacial failure process, an optimized inversion method suitable for rapid interface correction is proposed, a finite element model that can invert the entire process of interfacial loading failure is established, and an accurate degradation model of the interfacial bonding performance under corrosion is established. Summary of the Invention

[0011] The purpose of the present invention is to provide a method for correcting the mechanism model of the FRP-concrete interface in a corrosive environment. Through the inversion correction of the whole process, the optimization Latin hypercube sampling method and the gradient optimization method are used to correct the model respectively, compare the advantages and disadvantages of the two model correction methods, speed up the inversion correction process of the interfacial constitutive relationship, and further establish the refined correction and rapid correction methods of the interfacial constitutive model. By the method of model correction, the influence of test strain measurement errors on the test results is avoided, so the bonding-slip characteristics of the interface can be more truly reflected.

[0012] To achieve the above purpose, the present invention provides a method for correcting the mechanism model of the FRP-concrete interface in a corrosive environment, including the following steps:

[0013] S1. Extract the original data of relevant literatures. By retrieving relevant domestic and foreign literatures, determine the typical interfacial corrosion types, establish an experimental database with multiple influencing factors, and use image software to extract the slip curve and strain distribution curve of the test piece;

[0014] S2. Establish a finite element model of the FRP-concrete interface based on the interface bond-slip constitutive relation from the literature. Model the interface shear specimen using the Abaqus software. The constitutive relation of the concrete material adopts the plastic damage model, and the FRP fabric / plate is set as an isotropic and linear elastic material; use the cohesive element to simulate the interface bond-slip relationship, and the maximum nominal stress MAXS criterion is adopted for the initial damage;

[0015] S3. Invert and correct the interface constitutive model, call the Abaqus model and select the design variables and state variables, and construct the load-slip curve through the Calculator component with the state variables. Take the absolute area difference between the test curve and the initial simulation curve as the objective function for optimization, select a suitable optimization method and set constraints for iterative calculation, and finally obtain the inversion analysis results.

[0016] Preferably, in step S1, extract the original database of the literature related to the corrosion environment, select two cases of chloride corrosion and sulfate corrosion, and each case includes two corrosion methods of immersion and wet-dry cycle.

[0017] Preferably, in step S2, the finite element modeling analysis includes material constitutive input, dimension component establishment, mesh division, contact constraint and boundary condition setting, analysis step definition, and post-processing data extraction.

[0018] Preferably, the material constitutive input includes the concrete constitutive relation, the FRP constitutive relation, and the interface adhesive layer. The concrete constitutive relation uses the plastic damage model to describe the stress-strain relationship of the concrete, and uses isotropic damage elasticity combined with isotropic tensile and compressive plasticity to describe the plastic behavior of the concrete. The interface adhesive layer uses the cohesive model to simulate the interface force failure process and the overall response, and is defined by the bilinear traction-separation model Cohesive element.

[0019] Preferably, in step S3, the inversion and correction of the interface constitutive model includes model error analysis, design variable selection, method introduction, objective function, and determination of the best optimization method.

[0020] Preferably, in step S3, the design variables are the initial interface stiffness K, the maximum interface shear stress τ max , and the interface fracture energy G f , and the optimization method is the optimized Latin hypercube sampling method.

[0021] Preferably, in step S3, the calculation formula of the objective function F is as follows:

[0022]

[0023] Where: m i represents the load level number, yexp represents the measured slip value at a certain load level, y mol represents the simulated slip value at a certain load level.

[0024] The advantages and positive effects of the method for modifying the FRP-concrete interface mechanism model in the corrosion environment of the present invention are as follows:

[0025] 1. In view of the problem that the existing interface bonding performance model under corrosion action cannot accurately invert the interface failure process, the present invention proposes an optimized inversion method suitable for rapid modification of interface constitutive relations, and establishes a finite element model that can invert the entire process of interface loading failure.

[0026] 2. The present invention proposes an optimized inversion method for rapid modification and establishes an accurate degradation model of interface bonding performance under corrosion action.

[0027] 3. The present invention creates a library of interface specimens in corrosive environments by collecting literature, proposes a rapid optimization algorithm to achieve interface constitutive modification, and summarizes the mechanism of interface corrosion deterioration.

[0028] The technical solution of the present invention will be further described in detail below with reference to the drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flowchart of the method for modifying the FRP-concrete interface mechanism model in the corrosion environment of the present invention;

[0030] Figure 2 is an inversion flowchart of the present invention;

[0031] Figure 3 is a diagram of the bilinear traction-separation relationship of the interface of the present invention;

[0032] Figure 4 is a diagram of the interface model of the present invention, where A is a 5-mm unit grid and B is a 10-mm unit grid;

[0033] Figure 5 is a schematic diagram of the model boundary conditions of the present invention;

[0034] Figure 6 is a diagram of the error between the experimental data and the model data of the load-slip curve of the present invention;

[0035] Figure 7 is a characteristic curve diagram of load-slip of the present invention;

[0036] Figure 8 is an optimized Latin hypercube sampling diagram of the present invention;

[0037] Figure 9 is a schematic diagram of the objective function of the present invention;

[0038] Figure 10This is a comparison diagram of the optimization method of the present invention, where A is the optimization of the load-slip curve, and B is the modified constitutive model parameters;

[0039] Figure 11 This is a diagram of the results of modifying the load-slip curve of the present invention, where A is sulfate immersion, B is chloride salt immersion, C is chloride salt wet-dry cycle, and D is sulfate wet-dry cycle;

[0040] Figure 12 This is a constitutive model curve diagram of the present invention, where A is sulfate immersion, B is chloride salt immersion, C is chloride salt wet-dry cycle, and D is sulfate wet-dry cycle. Detailed implementation manners

[0041] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs.

[0043] Embodiment

[0044] A method for modifying the interface mechanism model of FRP-concrete in a corrosive environment includes the following steps:

[0045] S1. Extract the original data of relevant literatures. By searching relevant domestic and foreign literatures, determine several typical interface corrosion types, establish a test database with multiple influencing factors, and use image software to extract the slip curve and strain distribution curve of the test piece;

[0046] S2. Establish a finite element model of the FRP-concrete interface based on the interface bond-slip constitutive model in the literature. Use Abaqus software to model the interface shear test piece. The constitutive of the concrete material adopts the plastic damage model, and the FRP cloth / plate is set according to isotropic and linear elastic materials; Use cohesive elements to simulate the interface bond-slip relationship, and the initial damage adopts the maximum nominal stress MAXS criterion;

[0047] S3. Invert and modify the interface constitutive model, call the Abaqus model and select the design variables and state variables, and construct the load-slip curve through the Calculator component with the state variables. Take the absolute area difference between the test curve and the initial simulation curve as the objective function for optimization, select a suitable optimization method and set constraints for iterative calculation, and finally obtain the inversion analysis results.

[0048] I. Test database for corrosive environment

[0049] FRP-reinforced coastal bridges or concrete structural members in saline soil and salt lake areas are often affected by the corrosive environment. Under the corrosive environment, the interface peeling failure of FRP-concrete is more serious.

[0050] The corrosion environment database selects two cases of chloride corrosion and sulfate corrosion, and includes two corrosion methods of immersion and wet-dry cycling respectively.

[0051] Select the test data from the following literature:

[0052] Literature 1: Liu Shengwei. Research on the Degradation Law and Deterioration Mechanism of the Bonding Performance of CFRP-Concrete Interface under Sulfate Environment [D]. Lanzhou Jiaotong University, 2018;

[0053] Literature 2: Zhao Yingying. Experimental Study on the Bonding Fatigue Performance of CFRP-Concrete Interface under Marine Environment [D]. Qingdao University of Technology, 2015;

[0054] Literature 3: Fan Fuquan. Experimental Study on the Bonding Performance of CFRP-High Performance Concrete Interface under Marine Environment [D]. Qingdao University of Technology, 2015;

[0055] Literature 4: Dong Xiaosong. Experimental Study on the Bonding Performance of CFRP-Concrete Interface under Dry-Wet Cycling of Salt Solution [D]. Dalian University of Technology, 2014.

[0056] They are hereinafter referred to as Literature 1, Literature 2, Literature 3, and Literature 4.

[0057] There are a total of four working conditions, namely chloride immersion, chloride wet-dry cycling, sulfate immersion, and sulfate wet-dry cycling. The database parameters are derived from the bond-slip curves in the literature. For the literature without the bond-slip curves shown, only part of the data is extracted, and part of the data is calculated based on the Nakaba model. The parameters to be adjusted and corrected in the database are the stiffness K, the maximum shear stress τ max and the fracture energy G f See Table 1.

[0058] Table 1 Corrosion Environment Database

[0059]

[0060]

[0061] II. Finite Element Modeling

[0062] For the specimens in the above database, based on the information in the literature, finite element modeling of the specimens is carried out. The relevant material parameters for modeling are extracted from the literature, including the strength grade of concrete, specimen size, FRP size, elastic modulus and other parameters to complete the establishment of the model. The finite element model is based on Abaqus software, and its finite element analysis mainly includes: material constitutive input, size component establishment, mesh generation, contact constraint and boundary condition setting, analysis step definition, post-processing data extraction, etc.

[0063] (1) Material constitutive input

[0064] (1) Concrete constitutive relationship

[0065] The plastic damage model (hereinafter referred to as the CDP model) is adopted to describe the stress-strain relationship of concrete. The stress-strain relationships in tension and compression adopt the Guo Zhenhai model, which can better simulate the mechanical behavior of concrete.

[0066] When the tensile stress of concrete material does not exceed the tensile strength, the concrete material is in the elastic stage, and the stress-strain relationship is linear elastic; after exceeding the tensile strength, the concrete material begins to enter the plastic stage, the stress-strain relationship no longer maintains linear elasticity, the material appears fine damage cracks and cannot be restored, and the mechanical properties decline. When the concrete material is in compression, it maintains a linear elastic stress-strain relationship before cracking, enters the hardening stage after cracking, and enters the softening stage after exceeding the ultimate stress.

[0067] (2) FRP constitutive relationship

[0068] FRP fabrics / plates are anisotropic materials made of fibers. However, in the interface shear test, it is not stressed in the direction perpendicular to the fibers, and the tensile strength along the fiber length direction is much greater than that in the direction perpendicular to the fiber length. Therefore, FRP fabrics / plates can be simplified as isotropic materials in the modeling process.

[0069] When FRP fabrics / plates are stressed in the length direction (longitudinal direction), their stress-strain relationship is in a linear elastic state, and the elastic modulus and tensile strength of FRP fabrics / plates are much higher than those of concrete materials. In the interface shear test, FRP fabrics / plates are always in the elastic stage. Therefore, FRP fabrics / plates maintain linear elastic isotropic materials in the modeling process.

[0070] (3) Interface adhesive layer

[0071] The cohesive model (Cohesive element) is adopted to simulate the interface force failure process and overall response. The Cohesive element belongs to the traction-separation relationship and is defined by the bilinear traction-separation model Cohesive element. The characteristic values of the bilinear curve include the initial interface stiffness K, the maximum interface shear stress τ max and the interface fracture energy G f Three parameters, as Figure 3 shown.

[0072] Regarding the bilinear traction-separation model, Abaqus provides four damage initiation criteria, namely the maximum nominal stress criterion (Maxs), the maximum nominal strain criterion (Maxe), the quadratic nominal stress criterion (Quads), and the quadratic nominal strain criterion (Quade). The maximum nominal stress criterion (Maxs) is adopted as the damage initiation criterion.

[0073] After the material reaches the damage initiation criterion, an energy-based damage evolution method is adopted, and the BK fracture criterion is selected for the energy-based damage evolution.

[0074] For the bilinear traction-separation model, the elasticity, maximum nominal stress, and fracture energy of the material must be defined in the material properties, corresponding to K, τ max and G f in the bilinear curve respectively, and the elastic type is surface traction. The calculation results of 0-thickness Cohesive elements and Cohesive elements with thickness are the same, and the thickness of Cohesive elements does not affect the results.

[0075] (2) Establishment of dimensional components and element mesh generation

[0076] (1) Establishment of dimensional components

[0077] Based on the existing test data records, the interface shear specimens are modeled according to the actual dimensions of the components. For example, the size of the interface shear concrete specimen in Document 2 is 100mm×100mm×250mm, and the size of its modeled specimen is 50mm×100mm×250mm. Since the thickness of the interface adhesive layer does not affect the calculation results, it is uniformly set to 1mm, and the length and width are set according to the document respectively. The FRP layer is modeled according to the initial data in the document without simplification.

[0078] (2) Element mesh generation and mesh type selection

[0079] The calculation times of different specimens are different. At the same time, considering the convergence of the finite element calculation, for different specimens, the element sizes of the specimens are kept the same as much as possible, but there are still some cases where the element sizes do not match. Based on this, the element sizes will be appropriately adjusted, and the subsequent element sizes will be given in the form of a range.

[0080] The element type of concrete is an eight-node linear hexahedron element and the reduced integration C3D8R is used. Considering the calculation cost, the concrete elements should not be too small, so the mesh generation size is preferably 5mm to 10mm. The elements of FRP are the same as those of concrete, namely C3D8R, and the size division is the same. The adhesive layer uses the eight-node three-dimensional viscous element COH3D8, and the size of the viscous element is set to 2mm to 5mm. Figure 4In A, it is the 5mm unit grid division method. Figure 4 In B, it is the 10mm unit grid division method.

[0081] (3) Contact constraint and boundary condition setting

[0082] Finite element simulations were carried out on the specimens, but there were differences in the simulation methods, especially in the definition of contact and boundary conditions. Considering the subsequent correction based on the finite element model, the contact and boundary conditions of the specimens were uniformly defined to avoid errors in the load-slip curve caused by different boundary conditions and contacts in the future.

[0083] ① Contact constraint

[0084] The contact constraints for the interface shear specimens of FRP-concrete include three parts: concrete-interface adhesive layer, interface adhesive layer-FRP, and FRP-loading point.

[0085] For the concrete-interface adhesive layer, since the interface adhesive layer is modeled using solid elements rather than contact modeling, the constraint between the concrete and the interface adhesive layer is set as Tie. And according to the Abaqus user manual, the master surface of the Tie command is uniformly set as the upper surface of the concrete with a larger area, and the slave surface is set as the lower bottom surface of the interface adhesive layer.

[0086] For the interface adhesive layer-FRP, the Tie command is also adopted. The master surface is the FRP layer with greater stiffness, and the slave surface is the interface adhesive layer with smaller stiffness. For the FRP-loading point, the coupling constraint command is used. The center point on the side surface of the FRP is selected as the constraint control point and coupled with the side surface of the FRP. Based on the coupling constraint command, the RF-U curve of the loading point can be obtained.

[0087] ② Boundary condition setting

[0088] Boundary conditions refer to the edge constraint conditions of the specimens. More accurate boundary conditions are a necessary condition for accurate finite element simulations. The boundary condition setting method of fixing the upper and lower surfaces of the concrete completely and leaving the lower bottom surface in an unconstrained state is adopted, as Figure 5 shown. Because although only the translational motion in three directions is constrained in the test, rotation cannot occur during the test, so the upper and lower bottom surfaces of the concrete should be in a completely fixed state. The lower bottom surface of the concrete is not constrained in the test and is naturally in an unconstrained state.

[0089] (3) Definition of analysis steps and post-processing

[0090] (1) Analysis step setting

[0091] Due to the different degrees of convergence difficulty of different models, two analysis step methods are adopted for different models, namely the static general analysis step and the dynamic explicit analysis step. For models that are easy to converge, the static general analysis step is adopted, and its settings are based on experience. The analysis step time is 1.0, the minimum number of incremental steps is 10 - 15, the maximum number of incremental steps is 50000, and the initial incremental step is 0.01. The remaining settings remain default. For models that are difficult to converge, the dynamic explicit analysis step is used to simulate the static load test process. The analysis step time is set to 0.1, and mass scaling is applied to the model. The mass scaling coefficient can accelerate the calculation process, and the mass scaling coefficient can be appropriately increased within the range of calculation accuracy. The mass scaling coefficient of the present invention is 500.

[0092] III. Model Modification

[0093] (1) Model Error Analysis

[0094] The finite element modeling simulates the load - slip development of the specimen during the entire shear process to the greatest extent, but it is still found that there are errors between the results of the finite element model and the test results, and the errors show high discreteness. The errors have no directionality and exist within the range of the test data. The errors between the results of the finite element model and the test results are mainly manifested as deviations in the load - slip curve, and the deviations are mainly manifested in two aspects: the slopes of the curves during the entire process cannot fully match, and the characteristic bearing capacities in the three stages do not match.

[0095] Take Figure 6 as an example. Figure 6 Figure 16 shows the comparison of the load - slip curves of the indoor comparison specimen DB - 0 in Document 2. The dashed line is the test result and the solid line is the simulation result. It can be seen that there are certain errors in the changes of the slopes of the elastic section and the softening section of the load - slip curve. The slopes of each stage of the test curve are greater than those of the finite element model curve.

[0096] Based on the error analysis in Figure 6 , the characteristic points of the Figure 7 load - slip curve are analyzed for errors. The errors are calculated using the integral absolute error (IAE), as shown in Equation (1).

[0097]

[0098] In the formula: PM exp is the parameter test value; PM mol is the parameter model prediction value.

[0099] The data in the test database under the corrosion environment is statistically analyzed, and the IAE values are calculated according to the above method. The calculation results are shown in Table 3, where IAE A is the integral absolute error of the slope of the linear elastic section, is the integral absolute error of the maximum bearing capacity of the softening section. is the integral absolute error corresponding to the slip value at the maximum bearing capacity of the softening section, The calculation parameter is the integral absolute error of the maximum bearing capacity of the debonding and peeling section, The calculation parameter is the integral absolute error corresponding to the slip value at the maximum bearing capacity of the debonding and peeling section. According to the error calculation results, it is found that the integral absolute error of each control point from each literature source in the corrosion environment database is above 10%, and even the error of some control points is close to 20%. Among the five integral absolute error calculation values, the calculated value of IAE A is the largest, and the slope error of the linear elastic section is the largest. The error sources include not only the inherent error but also the calculation error. The coupling of the two errors leads to the largest calculated value of IAE A In the IAE evaluation, it is often considered that when its value is within 10%, the simulation effect is better. Therefore, comprehensively evaluating, the error between the simulated value and the test value of the specimens in the database is large, and the model needs to be corrected.

[0100] Table 3 Error analysis of the corrosion environment database

[0101]

[0102]

[0103] (2) Selection of design variables

[0104] The structural variable X is the interface constitutive model. In the finite element model, the interface constitutive model is a bilinear traction-separation model, and its model control parameters are the interface initial stiffness K, the interface maximum shear stress τ max and the interface fracture energy G f . Select the above three parameters as design variables, and their initial parameter settings refer to Table 1 of the database. The data not given in the database are set according to the calculation results of the Nakaba model.

[0105] (3) Optimized Latin hypercube sampling method

[0106] Isight can use the DOE component to conduct experimental design on the design variables, combine them within the value range of the design variables, so as to obtain several combinations for finite element calculation.

[0107] For three-factor design variables, the optimized Latin hypercube sampling method can freely select the number of sample combinations. According to the number of combinations, uniformly scatter points in its space to obtain sample combinations. This method stratifies (divides intervals) the original data, conducts random disordered sampling in each interval respectively, and combines them in disorder. In the case of a large number of sample points, this method can obtain relatively uniformly distributed spatial sample points. Figure 8This is the case of using the optimized Latin hypercube sampling method to generate samples of the interface constitutive model parameters for the specimen SW-A-0-180-0 in Document 1.

[0108] (4) Gradient optimization method

[0109] In Isight, in addition to using the DOE module to design the test variables, existing algorithms can also be used for objective optimization. The algorithm will automatically combine the variables based on its own operation logic and the set range of the variables and perform calculations.

[0110] The gradient optimization method makes the variables change along the negative gradient direction, that is:

[0111]

[0112] Thus, the function is optimized, and we can obtain:

[0113] f(x + Δx) < f(x) (3)

[0114] This process is continuously repeated until the function converges to a local minimum point. The characteristic of this method is to optimize through the gradient descent criterion, with a relatively fast optimization speed, but it may fall into a local optimal solution. For the correction of the three-factor multi-level model, the gradient optimization method has a short operation time while ensuring the accuracy, and it is the first choice. As a first-order derivative method, the gradient optimization method has better optimization effects on the premise of ensuring applicability, and the gradient optimization method is selected for subsequent comparison in the optimization algorithm.

[0115] (5) Objective function (F)

[0116]

[0117] In the formula: m i represents the load level number, y exp represents the measured slip value at a certain load level, and y mol represents the simulated slip value at a certain load level.

[0118] In Figure 1 and 2 in the technical roadmap, the objective function is a function of the structural response y exp and the finite element response y mol . In the interface pure shear test, the load-slip development curve is the only curve that can accurately reflect the overall response of the interface, while the strain distribution curve has discreteness in the longitudinal length direction of the FRP. Therefore, the load-slip curve is selected as the objective function.

[0119] The error between the load-slip curve calculated by the model and the test curve is constructed through the Datamatching component of the Isight software. AsFigure 9 As shown, this component uses the test curve as the Target, the finite element curve as the Simulation, and takes the absolute area difference S between the two curves as the objective function in the Data Comparison section.

[0120] (6) Method comparison

[0121] To compare two methods for model modification - the optimized Latin hypercube sampling method and the gradient optimization method, the two optimization methods are respectively used to modify the DB-0 specimen in Document II, and their results are compared. The initial parameter settings of this specimen are shown in Table 1 Corrosion Environment Test Database. The range of the interface stiffness K is set to 50 MPa / mm to 200 MPa / mm, and the maximum interface shear stress τ max The range is set to 2 MPa to 10 MPa, and the interface fracture energy G f The range is set to 0.2 MPa·mm to 4 MPa·mm. The number of iterations of the gradient optimization method and the number of samples of the optimized Latin hypercube sampling are set to 80, and other settings remain the software defaults. The comparison of the optimization processes is shown in Table 4. It can be seen from the table that the optimized Latin hypercube sampling method has a shorter operation time.

[0122] The model iteration optimization results are as Figure 10 shown. In Figure 10 A, it can be seen that when the original bond-slip constitutive model parameters are input into Abaqus, the deviation between the model and the test load-slip curve is large. The slope of the elastic section, the change of the slope of the softening section, and the ultimate bearing capacity of the model curve are all smaller than the test results. The results obtained by the two optimization methods are not the same. The correction effect based on the optimized Latin hypercube sampling method is better than that based on the gradient optimization method. The model modification based on the optimized Latin hypercube sampling method makes the simulated load-slip curve basically coincide with the test curve, makes the objective function F close to 0, and the ultimate bearing capacity and the slope of the elastic section of the interface can all coincide with the test.

[0123] Figure 10 In B, the bond-slip constitutive model parameters after modification are shown. It adopts a bilinear traction-separation relationship model, and the characteristic parameters include the initial interface stiffness K, the maximum interface shear stress τ max and the interface fracture energy G f . After the model modification, the three characteristic parameters of the interface have all changed. The maximum interface shear stress τ max and the interface fracture energy G f become larger, and the initial interface stiffness K remains almost unchanged.

[0124] Table 4 Comparison of optimized operation duration

[0125]

[0126] Based on the above analysis, it is considered that the optimized Latin hypercube sampling method is superior to the gradient optimization method in terms of operation duration, optimization effect, etc. Therefore, the optimized Latin hypercube sampling method is adopted in subsequent model corrections.

[0127] IV. Model Correction Results and Discussions

[0128] (1) Model Correction Results

[0129] Model corrections were performed on all specimens in the corrosion environment test database, and the model correction results of some specimens are shown for each influencing factor. The corrosion environment is divided into four categories of influencing factors, namely sulfate immersion, sulfate wet-dry cycle, chloride immersion, and chloride wet-dry cycle. Figure 11 In Figures A, B, C, and D, the correction results of each specimen under each influencing factor are shown. Since some literature did not give the load-slip curve, the slip value s(x) was obtained by integrating the FRP surface strain distribution curve using Equation (5), thereby obtaining the load-slip curve.

[0130]

[0131] In the formula: L is the bond length; ε(x) is the strain at a distance x from the free end. Assuming that the strain between two strain measurement points varies linearly, the slip value s at the i-th strain measurement point i+1 is:

[0132]

[0133] In the formula: The strain ε0 at the free end is defaulted to 0.

[0134] Figure 11 The corrected results of the load-slip curves under four corrosion influencing factors are respectively shown. The black curves in each subfigure are the test curves, and the gray curves are the curves after model correction. It can be seen that the curves after model correction basically coincide with the test curves in the linear elastic section and the softening section. The matching effect in the interface peeling section is slightly worse. This is mainly because the unevenness of the concrete interface leads to certain accidental errors in the load-slip curves obtained from the tests, and the finite element model correction cannot consider this accidental error. Therefore, the matching effect between the corrected curve and the test curve in the peeling section is slightly worse.

[0135] (2) Constitutive Parameter Correction Results

[0136] Literature research shows that compared with the corrosion time, the solution concentration (5% - 10%) has a smaller influence degree and can be ignored. Therefore, the solution concentration is not considered and unified fitting is carried out. For all specimens in the corrosion environment experiment database, the corrected constitutive parameters K, τ max , G f are obtained, as shown in Table 5.

[0137] Table 5 Revised Results of Corrosion Environment Test Database

[0138]

[0139]

[0140] Based on the revised results in Table 5, the characteristic points (s0, τ max ) and (s f , 0) of the bilinear traction-separation model are calculated through Equations (7) and (8), and the constitutive model curve is determined by the two characteristic points, as shown in Figure 12 .

[0141]

[0142]

[0143] Figure 12 This is the change trend of the constitutive model curve of the interface under the corrosion environment. Under the four corrosion influence factors, the parameters of the interface constitutive model all show a downward trend. Among them, the degree of decrease of the interface stiffness K is the lowest, and the slope of the linear rising section does not show a significant decrease, and it is the least affected by the corrosion environment. The next is the maximum interface shear stress τ max . The one most affected by the corrosion environment is the interface fracture energy G f . According to Equation (8), G f is affected by τ max and s f . The decrease of τ max and the increase of s f result in the largest degree of decrease in the interface fracture energy.

[0144] Figure 12 Figures A, B, C, and D in

[0145] Therefore, the present invention adopts the above-mentioned method for correcting the FRP-concrete interface mechanism model in the corrosion environment. Through the whole process of inversion correction, the optimization Latin hypercube sampling method and the gradient optimization method are used for model correction respectively. The advantages and disadvantages of the two model correction methods are compared to accelerate the constitutive inversion correction process of the interface, and further establish the refined correction and rapid correction methods for the interface constitutive model. By means of model correction, the influence of test errors on test results is avoided, so that the bond-slip relationship of the interface can be more truly reflected.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. Method for modifying FRP-concrete interface mechanism model in corrosive environment, characterized in that, It includes the following steps: S1. Extract the original data of relevant literature. By retrieving domestic and foreign literature on the corrosion of the FRP-concrete interface, determine the typical interface corrosion types, establish a test database with multiple influencing factors, and use image software to extract the slip curve and strain distribution curve of the specimens; S2. Based on the interface bond-slip constitutive model in the literature, establish a finite element model of the FRP-concrete interface. Use Abaqus software to model the interface shear specimens. The constitutive model of the concrete material adopts the plastic damage model, and the FRP cloth / plate is set according to isotropic and linear elastic materials; Use Cohesive elements to simulate the interface bond-slip relationship, and the initial damage adopts the maximum nominal stress MAXS criterion; S3. Invert and correct the interface constitutive model. Call the Abaqus model and select the design variables and state variables, and construct the load-slip curve through the Calculator component with the state variables. Take the absolute area difference between the test curve and the initial simulation curve as the objective function for optimization. Select a suitable optimization method and set constraints for iterative calculation to finally obtain the inversion analysis results.

2. The method for correcting the FRP-concrete interface mechanism model in a corrosion environment according to claim 1, characterized in that: In step S1, extract the original database of literature related to the corrosion environment, select two cases of chloride salt corrosion and sulfate corrosion, and each includes two corrosion methods of immersion and wet-dry cycle.

3. The method for correcting the corrosion environment FRP-concrete interface mechanism model according to claim 1, characterized in that: In step S2, the finite element model analysis includes material constitutive input, dimension component establishment, mesh generation, contact constraint and boundary condition setting, analysis step definition, and post-processing data extraction.

4. The method for correcting the FRP-concrete interface mechanism model in a corrosion environment according to claim 3, wherein: The material constitutive input includes the concrete constitutive relationship, the FRP constitutive relationship, and the interface adhesive layer. The concrete constitutive relationship uses the plastic damage model to describe the stress-strain relationship of concrete, and uses isotropic damage elasticity combined with isotropic tension and compression plasticity to describe the plastic behavior of concrete. The interface adhesive layer uses the cohesive force model to simulate the interface force failure process and overall response, and is defined by the bilinear traction-separation model Cohesive element.

5. The method for correcting the FRP-concrete interface mechanism model in a corrosion environment according to claim 1, characterized in that: In step S3, the inversion and correction of the interface constitutive model includes model error analysis, design variable selection, method introduction, objective function, and determination of the best optimization method.

6. The method for correcting the corrosion environment FRP-concrete interface mechanism model according to claim 1, characterized in that: In the step S3, the design variables are the initial interface stiffness K, the maximum interface shear stress τ max , the interface fracture energy G f , and the optimization method is the optimized Latin hypercube sampling method.

7. The method for modifying the FRP-concrete interface mechanism model in a corrosion environment according to claim 1, characterized in that: In step S3, the calculation formula of the objective function F is as follows: Where: m i represents the load level number, y exp represents the measured sliding value at a certain load level, y mol represents the simulated sliding value at a certain load level.

Images