A Risk Prediction Method for Shield-Type TBM Trucks Considering Time-Related Deformation of Surrounding Rock

By combining a nonlinear rheological constitutive model with a three-dimensional explicit finite difference platform, the problem of inaccurate prediction of jamming risk of shield-type TBMs under deep-buried, high-stress soft rock conditions was solved, achieving a complete description of the time-dependent deformation of the surrounding rock and an accurate assessment of jamming risk.

CN123116646APending Publication Date: 2026-11-17CHANGJIANG RIVER SCI RES INST CHANGJIANG WATER RESOURCES COMMISSION
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611617877.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies lack a complete description of the three stages of rock deformation—attenuation creep, steady-state creep, and accelerated creep—under conditions of deep-buried soft rock with high ground stress. Furthermore, the calculation of shield contact stress and frictional resistance is not substantially incorporated into the prediction of jamming risk. Geometric jamming and mechanical jamming criteria are calculated independently, resulting in jamming risk assessments that do not match engineering realities.

Method used

A nonlinear rheological constitutive model was constructed, including the improved nonlinear village mountain and Burgers model in series, combined with a three-dimensional explicit finite difference platform, to solve the age convergence displacement of the surrounding rock and the contact stress of the shield step by step in creep time, and to determine the risk of jamming by combining geometric and mechanical margins.

Benefits of technology

It achieves a complete description of the three-stage creep of rocks, the evolution of shield contact stress and frictional resistance over time, and the jamming risk assessment is consistent with engineering practice, providing a progressive and timely jamming risk assessment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116646A_ABST
    Figure CN123116646A_ABST
Patent Text Reader

Abstract

A method for predicting the jamming risk of a shield-type TBM considering the time-dependent deformation of surrounding rock is proposed, relating to the field of tunnel and underground engineering. This method constructs a rheological constitutive model that combines an improved nonlinear hillside model with a Burgers model, and completes three-dimensional difference and constitutive secondary development. In the TBM dynamic excavation model, the distribution of the time-dependent convergence displacement of the surrounding rock along the shield axis is solved step-by-step during creep. The contact initiation coordinates and contact segment are determined at the point where they are equal to the enlargement gap. The contact stress on the contact segment from the previous time step is applied as radial compressive stress to the surface of that section of surrounding rock. The total frictional resistance is obtained by integrating the contact stress on the contact segment and superimposing it with the shield's self-weight frictional resistance. The geometric margin is obtained from the contact initiation coordinates and shield length, and the mechanical margin is obtained from the rated thrust, total frictional resistance, and thrust required for normal continuous tunneling. The smaller of these is taken as the time-dependent jamming risk index. The remaining safe downtime is extrapolated based on its rate of change to determine the jamming risk.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of stability analysis technology for surrounding rock in tunnels and underground engineering, and specifically to a risk prediction method for shield-type TBM jacks that considers the time-dependent deformation of surrounding rock. Background Technology

[0002] Full-face tunnel boring machines (TBMs) are widely used in long-distance, deep-buried tunnel projects. Among them, shield tunneling machines (TBMs) are particularly sensitive to the deformation of the surrounding rock. Under geological conditions such as weak and fractured rock and high ground stress, shield jamming is a common problem affecting construction safety and progress. The deformation of deep surrounding rock under excavation disturbance often continues to develop over time. Especially in deep-buried soft rock sections with high ground stress, the compression deformation caused by the rheology of the surrounding rock accumulates with downtime. Therefore, the accuracy of the description of the time-dependent deformation of the surrounding rock directly affects the reliability of jamming mechanism analysis and risk prediction.

[0003] One existing approach employs a classical linear component-based rheological constitutive model. This model uses elastic, viscous, and plastic components as basic units, combining them in series and parallel to form Maxwellian, Kelvin, Burgers, Bingham, and Westburnian models to describe the instantaneous elastic deformation, decaying creep, and steady-state creep of rocks. Some 3D numerical platforms also have these creep constitutive models built-in for direct use. The component parameters of this type of model are taken as constants during calculation, and the strain rate does not change with the evolution of internal damage. Therefore, when the stress level approaches or exceeds the long-term strength of the rock and creep enters an accelerated phase, the described deformation development trend differs from the measured curves.

[0004] Another approach involves introducing nonlinear elements, such as unsteady viscosity coefficients that decay exponentially over time, into the classical model to characterize the sharp increase in strain rate during the accelerated creep stage. This improvement allows for the reproduction of the curve shape during the accelerated stage. However, because the nonlinear elements directly take over all viscoplastic strain exponentially after yielding, the decaying creep transition phase that objectively exists in the early stages of the plastic stage is not described in the model. Furthermore, the curve shape during the accelerated stage is highly sensitive to the value of the decay coefficient; even small changes in the parameter value can cause significant changes in long-term deformation.

[0005] Regarding risk prediction for machine jamming, one existing approach uses elastoplastic constitutive models to describe the surrounding rock. This involves numerically simulating different working conditions to obtain the shield contact stress and surrounding rock convergence, which are then used to construct samples and make risk assessments. In this approach, the surrounding rock deformation stabilizes after the excavation step is completed. The cumulative effect of continuous deformation of the surrounding rock during the downtime on the shield contact pressure and frictional resistance is not included in the calculation, resulting in a contact stress that does not change over time.

[0006] In the aforementioned numerical method for assessing the risk of machine jamming, the surrounding rock convergence and propulsion reserve are typically used as independent criteria: the former compares the surrounding rock convergence with the shield excavation gap to verify the design of the shield length and excavation volume; the latter compares the relationship between the sum of the shield frictional resistance and the tunneling thrust and the rated propulsion. These two criteria correspond to geometric jamming and mechanical immobility, respectively, and are not mutually transferable in the calculation: the calculation of convergence does not change the integration range of frictional resistance, and the shield contact stress corresponding to frictional resistance does not return to the surrounding rock surface to participate in the subsequent deformation solution. Correspondingly, the frictional resistance is integrated along the entire length of the shield, including the section in the front of the shield where the surrounding rock has not yet converged to contact the shield, resulting in an overestimation of the frictional resistance; while the surrounding rock surface is always in a free convergence state during the calculation, and the convergence during the shutdown period is also overestimated. These two deviations are in opposite directions and have unequal magnitudes, making their superposition difficult to predict in advance.

[0007] Therefore, existing technologies still have the following technical gaps under deep-buried, high-stress soft rock conditions: at the constitutive level, there is a lack of joint description of the transition between accelerated creep and viscoplastic decay creep, and the characterization of the three-stage creep is incomplete; at the numerical implementation level, there are few rheological constitutive calculation programs that can be directly called by a three-dimensional explicit finite difference platform and contain the above-mentioned improved mechanisms; at the prediction level, the aging deformation of the surrounding rock has not yet been substantially incorporated into the calculation process of shield contact stress and frictional resistance; and the two criteria of geometric jamming and mechanical jamming are calculated independently, the actual contact range between the shield and the surrounding rock that changes over time has not been identified, and the supporting effect of the shield on the surrounding rock has not been included in the solution of subsequent aging deformation, so there is still room for improvement between the judgment of jamming risk and engineering practice. Summary of the Invention

[0008] This invention proposes a method for predicting the risk of TBM jamming in shield-type TBMs that considers the time-dependent deformation of surrounding rock. The technical problem to be solved is: how to construct a nonlinear rheological constitutive model that can fully describe the three stages of rock deformation—decaying creep, steady-state creep, and accelerated creep—while also taking into account the viscoelastic hysteresis recovery after yielding. This constitutive model is implemented on a three-dimensional explicit finite difference platform using an incremental finite difference scheme that progresses with creep time steps, and then embedded into the numerical calculation of dynamic excavation of shield-type TBMs. In the solution of creep time steps, the time-dependent convergence displacement of the surrounding rock and the contact stress of the shield are mutually determined, so that geometric jamming and mechanical immobility are simultaneously considered in the same judgment, and the jamming risk metric and available downtime are given as they evolve over time.

[0009] To address the aforementioned technical problems, this invention provides a risk prediction method for shield-type TBM jacking machines that considers the time-dependent deformation of surrounding rock, comprising the following steps: Step S1: Construct a nonlinear rheological constitutive model consisting of an improved nonlinear village mountain and a Burgers model connected in series. The Burgers model consists of a Maxwell body and a Kelvin body connected in series. The improved nonlinear village mountain consists of a plastic switch and a series branch connected in parallel. The series branch consists of a Kelvin body and an exponentially unsteady Newtonian body whose viscosity decays with time connected in series. Step S2: Identify the rheological parameters of the constitutive model based on the rock graded loading creep test curve; Step S3: The three-dimensional creep equation of the constitutive model is derived into an incremental finite difference scheme, and the yield criterion on which the viscoplastic activation criterion and post-yield stress correction are based is determined. Based on this, a custom constitutive model is obtained through secondary development on a three-dimensional explicit finite difference platform. Step S4: In the shield-type TBM dynamic excavation numerical model, call the custom constitutive model to solve the axial distribution of the surrounding rock age convergence displacement and the shield contact stress step by step in the creep time step; Step S5: In each creep time step, take the axial coordinate at the point where the time-effect convergence displacement of the surrounding rock is equal to the shield excavation gap as the contact start coordinate, and take the section from the contact start coordinate to the tail end of the shield as the contact section; apply the shield contact stress on the contact section of the previous creep time step as the radial compressive stress boundary condition to the surface of the surrounding rock of the contact section. Step S6: Determine the total frictional resistance of the shield by integrating the shield contact stress along the shield surface on the contact section; determine the geometric margin by the contact start coordinate and the shield length; determine the mechanical margin by the TBM rated thrust, the total frictional resistance and the thrust required for normal continuous tunneling; take the smaller of the geometric margin and the mechanical margin to determine the time-related jamming risk index and determine the jamming risk.

[0010] Preferably, the one-dimensional creep constitutive equation of the constitutive model is a piecewise equation: When the initial stress is less than the yield strength: ; When the initial stress is greater than or equal to the yield strength, an improved nonlinear hill-mountain term is superimposed: ; in For the total strain of the model, For stress, For initial stress, For yield strength, Creep time, The attenuation coefficient is... and These represent the elastic modulus and viscosity coefficient of the Maxwell body in the Burgers model, respectively. and These represent the elastic modulus and viscosity coefficient of the Kelvin matrix in the Burgers model, respectively. and These represent the elastic modulus and viscosity coefficient of the Kelvin matrix in the improved nonlinear village hill, respectively. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. For switching functions, when Less than or equal to When taking 0, Greater than Time to take .

[0011] Preferably, the three-dimensional creep equation of the constitutive model is obtained by decomposing the stress tensor into a spherical tensor and a deviatoric tensor. The spherical strain is completed instantaneously and does not change with time, and the creep is caused only by the deviatoric stress tensor. The three-dimensional deviatoric strain components of the improved nonlinear village hill for: ; in For the deviatoric stress tensor, For yield deviatoric stress, The shear modulus of the Kelvin form in the improved nonlinear village hill is given. The viscosity coefficient of the Kelvin form in the improved nonlinear village hill is given. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. The attenuation coefficient is... Creep time; The yield function of the viscoplastic activation criterion is: , As the second invariant of stress deviator, For yield strength, when The viscoplastic deviatoric strain term of the improved nonlinear village hill is activated when it is greater than or equal to 0. When the value is less than 0, only viscoelastic deviatoric strain is updated.

[0012] Preferably, in step S2, the Boltzmann superposition principle is first used to convert the rock graded loading creep curves into separate loading creep curves under each stress level, and then the damped Levenberg-Marquardt algorithm is used to identify the rheological parameters. The iterative process is expressed as follows: ; in Let be the parameter vector for the s-th iteration. Let Hessian matrix be the objective function. The first gradient of the objective function. For the damping step size, It is an identity matrix.

[0013] Preferably, in step S3, the incremental finite difference scheme decomposes the total deviatoric strain increment into the sum of four deviatoric strain increments: instantaneous elastic body, Newtonian body, viscoelastic body and improved nonlinear body, and approximates the deviatoric stress and deviatoric strain within the time step with the mean of the new value and the old value at the time step. The deviatoric stress update formula for the entire constitutive model is obtained when the deviatoric stress has not reached the yield deviatoric stress: ; When the deviatoric stress reaches the yield deviatoric stress: ; in This is the deviatoric stress tensor, where the superscripts N and O represent the new and old values ​​at the time step, respectively. This represents the total partial strain increment. The deviatoric strain increment of the improved nonlinear village hill. This represents the deviatoric strain aging value for a viscoelastic body. and These are intermediate coefficients determined by the shear modulus, viscosity coefficient, and time step in the Kelvin model of the Burgers model. and The coefficients are determined by each elastic shear modulus, viscosity coefficient, time step, and the intermediate coefficients.

[0014] Preferably, the yield criterion upon which the post-yield stress correction is based is the Mohr-Coulomb criterion with tensile intercept. The Mohr-Coulomb criterion includes a shear instability criterion and a tensile instability criterion, wherein the shear instability criterion is... The tensile instability criterion is The boundary function between the shear plastic region and the tensile plastic region is: The uncorrelated shear potential function is The tension potential function is ,in and These are the maximum principal stress and the minimum principal stress, respectively. For cohesion, It is the internal friction angle. For shear expansion angle, For tensile strength, The coefficient is determined by the internal friction angle. The coefficient is determined by the shear dilatation angle. and These are used to determine the intermediate values ​​of the slope and intercept of the boundary function, respectively. The post-yield stress correction first calculates the test stress using viscoelasticity and then determines the principal stresses of the test stress. Next, it uses the shear instability criterion and the tensile instability criterion to determine whether the test stress exceeds the yield surface. Less than or equal to 0 and When the test stress is greater than or equal to 0, it is determined that the test stress has not exceeded the yield surface, and the test stress is taken as the final stress. When the test stress exceeds the yield surface and When the stress is less than 0, the test stress is determined to be located in the shear-plastic region. The test stress is then corrected back to the yield surface according to the shear potential function and the flow rule. The shear-plastic multiplier used for this correction... for: ; in and These are the maximum and minimum test principal stresses, respectively. and These represent the shear modulus and viscosity coefficient of the Kelvin matrix in the improved nonlinear village hill, respectively. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. The attenuation coefficient is... The step size of the creep time step; When the test stress exceeds the yield surface and When the value is greater than 0, it is determined that the test stress is located in the tensile plastic region, and the test stress is corrected back to the yield surface according to the tensile potential function and the flow law.

[0015] Preferably, in step S6, the total frictional resistance experienced by the shield... Calculate using the following formula: ; in For the shield radius, For shield length, The contact start coordinates are... The coefficient of friction between the shield and the surrounding rock. Coordinates along the shield axis Shield contact stress at the location, For the weight of the shield; The geometric margin is The mechanical margin is The time-sensitive card machine risk index is ; in The thrust required for normal continuous tunneling Rated thrust of the TBM The value is 1 when the shield has not yet made contact with the surrounding rock, and 0 when the contact section extends to the full length of the shield, i.e., the shield is completely covered by the surrounding rock; when A machine freeze is determined when the value is greater than or equal to 1; when Greater than or equal to the warning threshold And when it is less than 1, it is judged as a high risk of card malfunction and a card malfunction warning is issued; when Less than and Rate of change of creep time When greater than zero, press Calculate the remaining safe downtime, and determine the shutdown window or increase the shield excavation gap according to the remaining safe downtime; when When the value is less than or equal to zero, the risk of machine malfunction is determined to no longer increase.

[0016] Preferably, the constitutive secondary development in step S3 includes: declaring all parameters of the constitutive model according to the constitutive interface specification of the three-dimensional explicit finite difference platform, implementing member functions for initializing large deformation states and member functions for updating stress tensors and strain tensors step by step, compiling to generate a dynamic link library, loading the dynamic link library through the platform's plug-in configuration instructions, and allocating the custom constitutive model to the computational domain using constitutive assignment instructions.

[0017] Preferably, in step S4, the creep time step is converted into actual time according to the construction tunneling speed, and the creep time is advanced step by step according to the construction progress. The number of iterations of each creep time step is determined by dividing the sum of the tunneling period and the shutdown period by the step size of the creep time step. The step size of the creep time step is not greater than 1 / 10 of the ratio of the viscosity coefficient of the Kelvin in the Burgers model to the shear modulus of the Kelvin in the Burgers model. After the step size is halved, the relative deviations of the contact start coordinate, the total frictional resistance of the shield and the time-related jamming risk index do not exceed the preset tolerance.

[0018] Preferably, when When the second difference of the creep time is greater than zero, take the most recent creep time steps. Substituting the maximum value into the formula for calculating the remaining safe downtime, we obtain the remaining safe downtime.

[0019] The beneficial effects of the present invention include at least the following: 1) This invention comprises an improved nonlinear village hill body constructed by connecting an unsteady Newtonian body with a viscosity coefficient that decays exponentially over time and a Kelvin body in series. This is then connected in series with the Burgers model, allowing the viscoplastic strain after yielding to be shared by both the decaying creep transition segment and the accelerated creep segment. This preserves the transitional characteristic of decreasing strain rate in the initial stage of viscoplastic strain while reproducing the nonlinear steep increase in strain rate during the accelerated stage. Since the improved nonlinear village hill body does not participate in deformation when the plastic switch is not open, the Kelvin element still recovers with hysteresis after yielding as stress is removed, thus preserving the viscoelastic hysteresis recovery characteristic after yielding. Fitting results from triaxial graded loading creep tests on carbonaceous slate show that the coefficient of determination for each working condition is not less than 0.95, and the coefficient of determination for the unyielding stress level is not less than 0.98.

[0020] 2) This invention provides an incremental finite difference scheme for the three-dimensional creep equation, a yield criterion, and a stress correction method after yielding. Based on these, it completes the secondary constitutive development of a three-dimensional explicit finite difference platform, forming a constitutive program that can be loaded and allocated to the computational domain by the platform as a plug-in. This allows rheological calculations with the above-mentioned improved mechanisms to no longer rely on the existing form of the platform's built-in creep constitutive model. Numerical reproduction of creep test data for five rocks with different hardness—carbonaceous slate, mudstone, sandstone, argillaceous siltstone, and red mudstone—verifies that the calculation results of this constitutive program are consistent with the experimental curves.

[0021] 3) Since the creep time step corresponds to the actual time of the tunneling period and the shutdown period, the shield contact stress becomes a quantity that evolves with the construction process. The cumulative amplification of the shield contact pressure and frictional resistance due to the continuous time-dependent deformation of the surrounding rock during the shutdown period is reflected. The reserve relationship between the total frictional resistance and the thrust required for normal tunneling and the rated thrust changes with time. Therefore, the judgment of the risk of jamming is consistent with the gradual and time-dependent nature of shield jamming.

[0022] 4) Since the contact initiation coordinates are determined at the point where the distribution of the surrounding rock's aging convergence displacement along the shield axis equals the excavation gap at each creep time step, the integration range of frictional resistance shrinks to the actual contact section over time. Sections of the shield that have not yet contacted the surrounding rock are no longer included, and frictional resistance is no longer systematically amplified with the shield length. Furthermore, because the shield contact stress on the contact section in the previous time step is applied to the surrounding rock surface as radial compressive stress, the second invariant of the stress deviation in that section of the surrounding rock decreases accordingly, reducing the yield function value of the viscoplastic activation criterion. Consequently, the subsequent aging convergence rate slows down, and the surrounding rock convergence amount is no longer overestimated as calculated based on free convergence. If the integration range is only shrunk according to the convergence displacement without applying the contact stress back to the surrounding rock surface, the convergence amount develops according to free convergence, and the contact section expands too quickly towards the shield front, resulting in an overestimation of frictional resistance. If the contact stress is only applied back to the surrounding rock surface without shrinking the integration range, the frictional resistance is still integrated along the entire shield length, and the virtual contact stress of the uncontacted sections is included, again leading to an overestimation of frictional resistance. When both conditions are met, the contact segment range and the contact stress distribution are mutually conditional in the time-step solution. The values ​​of frictional resistance and convergence amount converge to a level that matches the actual contact state. The effect is not a simple superposition of the results obtained by implementing the two conditions separately.

[0023] 5) Since the geometric margin and mechanical margin are normalized by the shield length and rated thrust respectively, the smaller one can be taken under the same dimension. The resulting time-dependent jamming risk index will rise when either of the two jamming paths is close to exhaustion, avoiding the omission of geometric jamming when judging only by thrust reserve, and the omission of thrust exhaustion when judging only by contact section range. Furthermore, since the index changes continuously with creep time, its rate of change with time gives the rate at which thrust reserve and geometric clearance are swallowed up. The remaining safe downtime obtained by extrapolation can be directly used to determine the downtime window or decide whether to increase the excavation clearance. Attached Figure Description

[0024] Figure 1 The flowchart shows the risk prediction method for shield-type TBM chucks that takes into account the time-dependent deformation of surrounding rock according to the present invention. Figure 2 This is a schematic diagram of the component composition of the nonlinear rheological constitutive model of the present invention; Figure 3 A schematic diagram of the components of a traditional Burgers model; Figure 4 A schematic diagram of the components of a nonsteady viscoplastic body; Figure 5 A schematic diagram illustrating the components of an improved nonlinear village hill. Figure 6 The creep curves for triaxial graded loading tests on carbonaceous slate; Figure 7 Creep curves for various parameter values ​​in the rheological parameter sensitivity analysis; Figure 8 A schematic diagram of the Mohr-Coulomb yield criterion with tensile intercept; Figure 9 A schematic diagram of the Mohr-Coulomb flow law; Figure 10 This is a schematic diagram of the numerical model for a triaxial creep test; Figure 11 The vertical displacement contour plot is for the numerical model of the triaxial creep test. Figure 12 A comparison curve showing the numerical calculation and experimental values ​​of carbonaceous slate; Figure 13 A schematic diagram for calculating the contact and frictional resistance between the shield and the surrounding rock; Figure 14 This is a schematic diagram illustrating the process of determining the initial contact coordinates and the contact segment. Figure 15 The curve showing the comparison between the numerical calculation values ​​and experimental values ​​for mudstone. Figure 16 A comparison curve showing the numerical calculation and experimental values ​​for sandstone; Figure 17 The curve showing the comparison between the numerical calculation and experimental values ​​of silty mudstone; Figure 18 This is a comparison curve between the numerical calculation values ​​and experimental values ​​of red mudstone. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0026] The overall process of this invention is as follows: Figure 1As shown, the process includes six steps: constructing a nonlinear rheological constitutive model considering three-stage creep; identifying rheological parameters based on graded loading creep tests on rocks; deriving the three-dimensional creep equation into an incremental finite difference scheme and determining the viscoplastic activation criterion and the yield criterion upon which post-yield stress correction is based, thus completing the secondary development of the constitutive model; calling a custom constitutive model in the dynamic excavation numerical model of a shield-type TBM to solve for the aging convergence displacement and shield contact stress step by step in creep; determining the contact segment in each creep time step and applying the contact stress of the previous time step back to the surrounding rock surface of that segment; and calculating the total frictional resistance by integrating the contact stress on the contact segment and determining the aging jamming risk index and remaining safe downtime based on the geometric and mechanical margins. The specific lithology, numerical platform, and parameter values ​​mentioned are examples and do not constitute a limitation on the scope of protection.

[0027] Step S1: Construct a nonlinear rheological constitutive model consisting of an improved nonlinear village hill and a Burgers model connected in series. The Burgers model consists of a Maxwell body and a Kelvin body connected in series. The improved nonlinear village hill consists of a plastic switch and a series branch connected in parallel. The series branch consists of a Kelvin body and an exponentially unsteady Newtonian body whose viscosity decays with time connected in series.

[0028] This embodiment presents the component composition of a nonlinear rheological constitutive model and its one-dimensional and three-dimensional constitutive equations. The component composition of the constitutive model is as follows: Figure 2 As shown, it is composed of Maxwell bodies, Kelvin bodies, and improved nonlinear hill bodies connected in series. The Maxwell bodies are composed of bodies with an elastic modulus of... The elastic element and viscosity coefficient are It is composed of viscous elements connected in series, and the Kelvin body is made of elastic modulus of The elastic element and viscosity coefficient are The viscous elements are connected in parallel, improving the nonlinearity of the village mountain with a yield strength of The plastic switch is connected in parallel with a series branch, and the elastic modulus of the series branch is... The elastic element and viscosity coefficient are After the viscous elements are connected in parallel to form a Kelvin body, they are connected in series with an unsteady Newtonian body whose viscosity decays exponentially over time and whose decay coefficient is A.

[0029] The basic components constituting the above model follow the following constitutive relation: When an elastic element is subjected to force, it produces instantaneous elastic strain independent of time, and its constitutive relation is: In the formula For stress, In response, Let be the elastic modulus of the elastic element; for the viscous element, there is no instantaneous deformation, its strain increases with time and does not recover after unloading, and its constitutive relation is: In the formula, η is the viscosity coefficient of the viscous element. For time; plastic elements do not produce strain when the stress is less than the yield strength, and strain develops unrestricted when the stress reaches the yield strength. The components of the traditional Burgers model are as follows: Figure 3 As shown, it is composed of Maxwellian and Kelvinian structures connected in series, and can describe instantaneous elasticity, decaying creep, and steady-state creep. Its creep equation is: ; In the formula For initial stress, Creep time, and These are the elastic modulus and viscosity coefficient of the Maxwell body, respectively. and These represent the elastic modulus and viscosity coefficient of the Kelvin matrix, respectively. Since the above component parameters are all constants and do not change with time, the strain rate described by this model does not increase with the evolution of internal damage, and therefore cannot describe the accelerated creep stage.

[0030] To characterize accelerated creep, this invention replaces the Newtonian elements in the Bingham body with a nonsteady Newtonian body whose viscosity coefficient decays exponentially with time, resulting in a nonsteady viscoplastic body. Its component composition is as follows: Figure 4 As shown, the yield strength is The plastic switch is connected in parallel with the unsteady Newtonian body, and its viscosity coefficient is: ; In the formula Here, A is the initial value of the viscosity coefficient, and A is the attenuation coefficient. This element only activates after the stress exceeds the yield strength, and its creep differential equation is: Solving by integration, we get: ; In the formula For yield strength, For switching functions, when Less than or equal to hour Take 0, when Greater than hour Pick As can be seen from the above formula, the attenuation coefficient A determines the form of the accelerated creep curve, and its value has a significant impact on the long-term deformation.

[0031] A simple unsteady viscoplastic body exponentially takes over all viscoplastic strain after yielding, and the decaying creep transition phase in the early plastic stage of the rock is not reflected. Therefore, this invention connects an unsteady Newtonian body with a Kelvin body in series to form an improved nonlinear mountain body, the components of which are as follows: Figure 5As shown, the plastic switch is connected in parallel with a branch consisting of a Kelvin element and a non-steady Newtonian element connected in series; its structure is similar to that of the Cunshan body, retaining both the decaying creep transition and viscoelastic hysteresis recovery characteristics described by the Kelvin element in the early viscoplastic stage and the accelerated creep characterized by an exponential nonlinear element. According to the creep superposition principle, the improved nonlinear Cunshan body's creep equation is: ; In the formula and These are to improve the elastic modulus and viscosity coefficient of the Kelvin element in the nonlinear village hill. The initial value of the viscosity coefficient of its unsteady Newtonian body.

[0032] By cascading the improved nonlinear village hill model with the traditional Burgers model, a nonlinear rheological constitutive model is obtained. Each element in the model describes a different creep stage: the Maxwellian solid describes instantaneous deformation and steady-state creep, the Kelvin solid describes decaying creep, and the improved nonlinear village hill model describes viscoplastic deformation after yielding. Less than The plastic switch is not activated, and the transient elastomer satisfies the condition. Newtonian body satisfies viscoelastic materials satisfy In the formula , and These are the component stresses borne by transient elastic bodies, Newtonian bodies, and viscoelastic bodies, respectively. , and These are the component strains generated by the above three factors; each component strain satisfies the following relationship with the initial stress: and Solving the equations simultaneously yields the one-dimensional creep equation: ; when Greater than or equal to When the plastic switch is activated, the component strain of the nonlinear village hill is improved. Superimposed on the above equation, by the superposition principle, we get: ; Therefore, the one-dimensional creep constitutive equation of the constitutive model is piecewise: when Less than When taking the previous formula, Greater than or equal to The latter form is used.

[0033] Under three-dimensional stress, assuming the volumetric deformation of the rock is instantaneous and elastic, remaining constant over time, the stress spherical tensor does not affect creep, only the deviatoric tensor does, the Poisson's ratio is constant during creep, and all parameters are constant except for the viscosity coefficient of the improved nonlinear mountain body, which varies with time. The stress and strain tensors are decomposed into a spherical tensor and a deviatoric tensor, respectively. and ,in For stress tensor, For strain tensor, The first invariant of the strain tensor is volume strain. For spherical stress, For the deviatoric stress tensor, For the partial strain tensor, Let Kronecker function be the parameter. The three-dimensional creep constitutive relations of each element are as follows: Transient elastic body satisfies and ,in and In the formula For spherical strain, Poisson's ratio, Instantaneous shear modulus, Bulk modulus; Newtonian body satisfaction In the formula The viscosity coefficient of a Newtonian solid; viscoelastic materials satisfy Solving for the given information, we can obtain the following results: In the formula and These represent the shear modulus and viscosity coefficient of the Kelvin model, respectively. In the above formulas, the superscripts 1, 2, and 3 for deviatoric stress and deviatoric strain correspond to the instantaneous elastic body, the Newtonian body, and the viscoelastic body, respectively.

[0034] The improved nonlinear village hill is activated after the deviatoric stress reaches the yield deviatoric stress, and its three-dimensional deviatoric strain components are: ; In the formula, the superscript 4 corresponds to the improved nonlinear village mountain. For yield deviatoric stress, To improve the shear modulus of the Kelvin element in the nonlinear village hill, an extended over-stress theory is adopted to describe the viscoplastic strain rate of the element. The viscoplastic strain rate is taken as the ratio of the over-stress function to the viscosity coefficient multiplied by the partial derivative of the potential function with respect to the force. Its form is consistent with the classical plastic flow law, and the scaling factor is the ratio of the over-stress function to the viscosity coefficient. The over-stress function is zero when the yield function is less than zero, and takes the power function form of the yield function when the yield function is greater than or equal to zero. In this embodiment, the power exponent is 1, and the initial value of the yield function is 1. The yield function of the improved viscoplastic activation criterion for the nonlinear village hill is: ; In the formula It is the second invariant of stress deviator. When The viscoplastic deviatoric strain term of the improved nonlinear village hill is activated when it is greater than or equal to zero. When the value is less than zero, only viscoelastic deviatoric strain is updated. This criterion is at a different level from the Mohr-Coulomb yield criterion on which the post-yield stress correction is based. The former determines whether the viscoplastic deviatoric strain term participates in the deviatoric strain decomposition of this time step, while the latter determines whether the trial stress needs to be corrected back to the yield surface.

[0035] Therefore, the piecewise form of the three-dimensional constitutive equation is obtained: when When less than zero ;when When the value is greater than or equal to zero, the improved viscoplastic term of the nonlinear village mountain is superimposed on the above equation. Under the constant confining pressure triaxial stress state, substituting the spherical stress and deviatoric stress into the above equation, we can obtain the constant confining pressure triaxial creep equation characterized by shear modulus and bulk modulus, whose structure is the same as the piecewise form of the one-dimensional creep equation.

[0036] Step S2: Identify the rheological parameters of the constitutive model based on the creep test curves of rock graded loading.

[0037] Triaxial creep tests on rocks typically employ a graded loading method to apply axial stress, resulting in curves that represent the cumulative effect of each stress level. Therefore, this invention first uses the Boltzmann superposition principle to convert the graded loading creep curve into separate loading creep curves at each stress level. This involves subtracting the creep contribution of each previous stress level at that moment from the measured strain at a given stress level, yielding the creep curve when that stress level acts alone. Subsequently, using the separate loading curves as the target, the damped Levenberg-Marquardt algorithm is used to invert the rheological parameters of the constitutive model. The iterative expression is: ; In the formula For the first The parameter vector for the next iteration. Let Hessian matrix be the objective function. The first gradient of the objective function. For the damping step size, The identity matrix is ​​used. This algorithm uses a small damping step size when the objective function decreases rapidly, behaving similarly to the Gauss-Newton method; and uses a larger damping step size when the decrease is hindered, behaving similarly to the gradient descent method. Therefore, it exhibits good convergence for identification problems involving rheological parameters with significantly different dimensions and ill-conditioned objective functions. During identification, only the curve at the unyielding stress level is calibrated. , , , and Five parameters were used to further calibrate the curve for the yield stress level while keeping the aforementioned parameters fixed. , , With the four parameters A.

[0038] Taking the triaxial graded loading creep test of carbonaceous slate as an example, its graded loading creep test curve is as follows: Figure 6 As shown, confining pressures were 0 MPa, 20 MPa, and 40 MPa, with axial stress applied in five progressively increasing increments under each confining pressure. When the confining pressure was 0 MPa, the axial stresses at each increment were 18.68 MPa, 23.35 MPa, 28.02 MPa, 32.69 MPa, and 37.36 MPa, respectively; when the confining pressure was 20 MPa, they were 33.21 MPa, 41.51 MPa, 49.81 MPa, 58.11 MPa, and 66.42 MPa, respectively; and when the confining pressure was 40 MPa, they were 47.62 MPa, 59.53 MPa, 71.43 MPa, 83.34 MPa, and 95.24 MPa, respectively. The results of the loading curve identification for each increment are as follows: When the confining pressure was 0 MPa and the axial stress was 37.36 MPa, Take 6.27 GPa, Take 2.728 GPa, Take 3.83 GPa·d, Take 28.1 GPa, Take 0.243 GPa·d, Take 3.5 GPa, Take 0.738 GPa·d, Taking 72.6 GPa·d, A = 0.582, and the coefficient of determination = 0.99; when the confining pressure is 20 MPa and the axial stress is 66.42 MPa, Take 5.11 GPa, Take 2.223 GPa, Take 1250 GPa·d, Take 8.88 GPa, Take 0.376 GPa·d, Take 8.5 GPa, Take 3.06 GPa·d, Taking 614 GPa·d, A = 0.266, and the coefficient of determination = 0.96; when the confining pressure is 40 MPa and the axial stress is 95.24 MPa, Take 6.31 GPa, Take 2.746 GPa, Take 4000 GPa·d, Take 5.72 GPa, Take 4000 GPa·d, Take 2.61 GPa, Take 0.281 GPa·d, With a value of 39200 GPa·d and A of 0.167, the coefficient of determination is 0.95. The coefficient of determination for each working condition is not less than 0.95, and not less than 0.98 at the unyielding stress level, indicating that the constitutive model can simultaneously fit the three creep stages of decay, steady state, and acceleration.

[0039] In the single-parameter sensitivity analysis of the parameters, The following values ​​were taken sequentially: 2 GPa, 5 GPa, 8 GPa, 11 GPa, 14 GPa, and 17 GPa. The following values ​​were taken sequentially: 20 GPa·h, 50 GPa·h, 80 GPa·h, 110 GPa·h, 140 GPa·h, and 170 GPa·h. The creep curves were obtained for the following parameters: 5000 GPa·h, 10000 GPa·h, 15000 GPa·h, 20000 GPa·h, 25000 GPa·h, and 30000 GPa·h, with A values ​​of 0.10, 0.15, 0.20, 0.25, 0.30, and 0.35 respectively. Figure 7 As shown. The analysis results indicate that the shear modulus The deformation amplitude and transition characteristics of controlled plastic stage decay creep were studied. When the pressure increased from 2 GPa to 17 GPa, the final strain decreased from 0.627% to 0.571%, a decrease of 8.9%. Viscosity coefficient... Controlling the evolution rate of decay creep, the final strain decreased from 0.589% to 0.581% when the rate increased from 20 GPa·h to 170 GPa·h, a decrease of 1.4%; viscosity coefficient Controlling the long-term total rheological deformation and the development of accelerated creep, the final strain decreased from 0.901% to 0.468% when the coefficient increased from 5000 GPa·h to 30000 GPa·h, a decrease of 48.1%. The attenuation coefficient A controlled the development of accelerated creep and the long-term total deformation; when it increased from 0.10 to 0.35, the final strain increased from 0.385% to 2.297%, an increase of 496.6%. Therefore, A and [other factors] should be prioritized in engineering calibration. , and It can take values ​​within the normal range.

[0040] Step S3: The three-dimensional creep equation of the constitutive model is derived into an incremental finite difference scheme. The viscoplastic activation criterion and the yield criterion on which the stress correction after yielding is based are determined. Based on this, a custom constitutive model is obtained through secondary development on a three-dimensional explicit finite difference platform.

[0041] Specifically, this embodiment discretizes the continuous solution domain into regular grid nodes using the finite difference method, approximates the differential with difference, and transforms the partial differential equations into a system of discrete node algebraic equations for iterative solution. Based on explicit time integration and incremental stepping, it can accurately characterize time-dependent deformation. The three-dimensional explicit finite difference platform is developed based on C++ object-oriented programming. Its constitutive module and solver are independent of each other. Custom constitutive algorithms can be embedded through dynamic link libraries, and the source files of the basic constitutive model are made public for theoretical extensions on models with similar mechanical mechanisms. This embodiment will not elaborate further on these points.

[0042] The differential scheme is established starting with the decomposition of the total partial strain increment. The total partial strain increment within a time step is decomposed into the sum of the partial strain increments of four parts: the instantaneous elastic body, the Newtonian body, the viscoelastic body, and the improved nonlinear body. Each component bears the same eccentric stress, that is To achieve second-order accuracy in the difference scheme, the mean of the new and old values ​​at each time step is used to approximate the deviatoric stress and deviatoric strain within that time step. Specifically, the mean of the new and old values ​​at each time step is taken as follows: and The superscripts N and O represent the new and old values ​​at the time step, respectively. Therefore, the instantaneous deviatoric strain increment of the elastic body satisfies... The partial strain increment of a Newtonian body satisfies The viscoelastic body satisfies a viscoelastic constitutive relation within a time step, expressed in terms of the mean values ​​of deviatoric stress and deviatoric strain, where This is the creep time step.

[0043] Rearranging the above equation for viscoelastic bodies, we obtain an updated equation for their deviatoric strain: ; in: ; ; In the formula and For the shear modulus of Kelvin viscosity coefficient With creep time step Determined intermediate coefficients.

[0044] The improved nonlinear village hill is activated after yielding, and its deviatoric strain increment within the time step is determined by the yield function and the potential function according to the flow law, that is: ; In the formula Let be the shear yield function. Let be the shear potential function, and the yield function in the above formula is taken as its positive part, that is, when the yield function is negative, this term is zero.

[0045] Substituting the deviatoric strain increments of each component into the decomposition equation of the total deviatoric strain increment, and rearranging, we obtain the deviatoric stress update equation for the entire constitutive model. When the deviatoric stress reaches the yield deviatoric stress, i.e., the plastic switch is turned on: ; When the deviatoric stress does not reach the yield deviatoric stress, and only the transient elastic body, Newtonian body, and viscoelastic body are at play: ; in: ; ; In the formula and These are coefficients determined by the shear modulus, viscosity coefficient, creep time step, and intermediate coefficient of each component. The spherical stress is then calculated according to... Update, in the formula The volumetric strain increment within the time step. This represents the increase in viscoplastic strain caused by shear dilatation after yielding; the latter is zero when the plastic switch is not open, at which point the spherical strain is instantaneous and does not change with time, consistent with the basic assumptions of the three-dimensional constitutive model. The increase in viscoplastic strain is obtained by adding the increases in viscoplastic principal strain in the three principal directions: in the shear plastic region, the sum of the partial derivatives of the shear potential function with respect to the three principal stresses is... Therefore In the tensile plastic region, the sum of the partial derivatives of the tensile potential function with respect to the three principal stresses is: Therefore In the formula and These are the shear and tensile plastic multipliers, respectively; the spherical stress update formula... Two notations that use the same amount as the strain increment in viscoplastic bodies in stress modification.

[0046] The yield criterion used for post-yield stress correction is the Mohr-Coulomb criterion with tensile intercept, such as... Figure 8 As shown, it is bounded by the shear yield surface and the tensile yield surface within the principal stress plane. The shear instability criterion and the tensile instability criterion are respectively: ; ; in , and These are the maximum principal stress and the minimum principal stress, respectively. For cohesion, It is the internal friction angle. Let be the tensile strength, and the maximum value of the tensile strength is . The uncorrelated shear potential function is ,in , The dilatation angle is given; the tensile potential function takes the correlation form. The boundary function between the shear plastic region and the tensile plastic region in the principal stress space is: ,in and These are the intermediate values ​​of the slope and intercept of the boundary function, respectively. For example... Figure 9 As shown, when When the stress is less than zero, the stress point is located in the shear plastic region and is corrected according to the shear flow law; when When the stress is greater than zero, the stress point is located in the tensile plastic region and is corrected according to the tension flow law.

[0047] The solution process is as follows: First, assume viscoelasticity to obtain the test stress, decompose it into principal stresses and determine the principal directions, then judge according to the above criteria. If it does not exceed the yield surface, the test stress is taken as the final stress; if it exceeds the yield surface, the test stress is corrected back to the yield surface according to the corresponding flow law. The correction is carried out in the principal stress space, and the correction amounts for deviatoric stress and spherical stress are respectively... and superscript in the formula Indicates the trial value obtained according to viscoelasticity, superscript Indicates the viscoplastic component, and These are the viscoplastic deviatoric strain increment and the viscoplastic bulk strain increment, respectively. Expressed in principal stress form, they are the trial principal stress minus the correction determined by the viscoplastic principal strain increment and the elastic modulus, where the elastic modulus is... and .

[0048] In the shear plastic region, the partial derivatives of the shear potential function with respect to the three principal stresses are: , and The viscoplastic principal strain increment is obtained by multiplying the plastic multiplier and the partial derivative according to the flow law, i.e. 、 、 Substituting this into the elastic relationship in the principal stress space, we obtain the corrected principal stresses as follows: ; ; ; The shear plastic multiplier is not solved by the uniformity condition, but is directly given by the over-limit stress theory. It corresponds to the viscoplastic deviatoric strain component of the improved nonlinear village hill and evolves with creep time step. Its expression is: ; In the formula and These represent the shear modulus and viscosity coefficient of the Kelvin form in the improved nonlinear village hill. To improve the initial value of the viscosity coefficient of an unsteady Newtonian body in a nonlinear village hill, The attenuation coefficient is... Let be the creep time step size; the numerator is the positive part of the value obtained by substituting the trial stress into the shear instability criterion. In the tensile plastic region, the plasticity multiplier is... The corrected principal stress is 、 and After correction, the principal stresses are transformed back to the global coordinate system along the principal directions to obtain the final stress tensor for that time step. Since the plastic part of this model consists of a Kelvin element and an improved nonlinear element connected in series, and both share the same yield function, only one stress correction is needed for each time step in the numerical iteration.

[0049] Based on the aforementioned difference scheme and stress correction process, secondary development of the constitutive model was completed on a three-dimensional explicit finite difference platform. Using the header and source files of the platform's built-in creep constitutive model Cvisc as a model, a custom constitutive model was implemented in a C++ integrated development environment, named Deemo in this embodiment. The header file declares all parameters of the model, including bulk modulus (bulk_), Maxwell body shear modulus (Mshear_), Maxwell body viscosity (Mviscosity1_), Kelvin body shear modulus (Kshear1_), Kelvin body viscosity (Kviscosity1_), six Kelvin body deviatoric strain components (Mekd_), plastic Kelvin body shear modulus (Kshear2_), plastic Kelvin body viscosity (Kviscosity2_), plastic modified Newtonian body viscosity (Mviscosity2_), attenuation coefficient (A_), cohesion (cohension_), internal friction angle (friction_), dilation angle (dilation_), tensile strength (tension_), shear plastic deformation (sHP_), and tensile plastic deformation (tHP_), and adds the creep identifier (isCreep) member function to declare the constitutive model as a creep constitutive model. The base class defines nine core member functions: `getName` returns the unique identifier of the constitutive, `getMinorVersion` returns the minor version number, `getProperties` enumerates parameter keywords, `getStates` manages the yield state, `getProperty` and `setProperty` respectively handle parameter value retrieval, parameter type conversion, and numerical mapping, `copy` performs deep copying and assigns custom members, `initialize` initializes the large deformation state and triggers core calculations, and `run` iteratively updates the stress tensor and strain tensor at each time step. The `run` function is implemented using the aforementioned segmented update method: first, it updates the trial deviatoric stress and spherical stress using viscoelasticity; then, it uses the yield function to determine if the plastic region has been entered. If so, it performs stress correction according to shear or tensile flow rules and accumulates shear and tensile plastic deformations. The source files integrate proprietary header files and platform-provided state, transformation, version, and algorithm header files, are registered via `DllMain`, and generate a dynamic link library using `Release` and `x64` configurations. During calculation, the dynamic link library is first loaded using plugin configuration instructions, and then the custom constitutive is assigned to the computational region using constitutive assignment instructions.

[0050] To verify the correctness of this constitutive procedure, a system is established as follows: Figure 10 The cylindrical specimen numerical model shown has a diameter of 50 mm and a height of 100 mm. The model is Z-axis along the height direction, with a total of 2090 nodes and 2000 elements. Axial stress is applied at the top, confining pressure is applied circumferentially, and vertical constraint is applied at the bottom. The Deemo constitutive model is invoked using plug-in configuration commands and constitutive assignment commands to calculate the vertical displacement at the specimen center and convert it into axial percentage strain. The resulting vertical displacement contour plot is shown below. Figure 11 As shown. A comparison between the calculated creep curve and the creep test curve of carbonaceous slate under triaxial graded loading is presented. Figure 12 As shown: the attenuation segment and steady-state segment of the curve under low stress level are partially described by the Burgers model, while the acceleration segment of the curve under high stress level is reproduced by the improved nonlinear village mountain model. The two are in good agreement at all stress levels.

[0051] Step S4: In the shield-type TBM dynamic excavation numerical model, call the custom constitutive model and solve the axial distribution of the age convergence displacement of the surrounding rock and the contact stress of the shield step by step in the creep time step.

[0052] This embodiment presents an implementation method for embedding a custom constitutive model into the dynamic excavation numerical calculation of a shield-type TBM and using this model to determine the risk of machine jamming. The method is applied to a soft rock section of a deep-buried, long-distance water diversion tunnel, where the surrounding rock is mainly Class IV and Class V soft rock, with great burial depth and high ground stress. During the excavation process, there are conditions of large compression deformation and shield jamming.

[0053] After completing the secondary development of the constitutive model, a numerical model of dynamic excavation of a shield-type TBM was established in a three-dimensional explicit finite difference platform. The calculation range of the model along the tunnel axis and transverse direction was taken to be three to five times the tunnel excavation diameter to weaken the boundary effect. Normal displacement constraints were applied around the model, fixed constraints were applied at the bottom, and equivalent stress of the overlying rock mass self-weight was applied at the top according to the burial depth. The surrounding rock elements were given the aforementioned custom constitutive model. The shield and segments were simulated as linear elastic structural elements. A contact surface was set between the outer surface of the shield and the surrounding rock to transfer normal contact stress and tangential frictional resistance. Excavation was carried out ring by ring along the axis. The shield element at the corresponding position was activated after each ring of excavation was completed.

[0054] Step S5: In each creep time step, take the axial coordinate at the point where the aging convergence displacement of the surrounding rock is equal to the shield excavation gap as the contact start coordinate, and take the section from the contact start coordinate to the tail end of the shield as the contact section; apply the shield contact stress on the contact section of the previous creep time step as the radial compressive stress boundary condition to the surface of the surrounding rock of the contact section.

[0055] like Figure 13 As shown, the outer surface of the shield 201 contacts the surrounding rock 202, and the cutterhead 203 is located at the front end of the shield. Let the axial coordinate of the shield be... Its origin is located at the rear edge of the cutterhead, pointing positively towards the tail end of the shield. The radial gap between the outer surface of the shield and the excavation profile is the widening gap. After the excavation face is passed, the surrounding rock converges into the tunnel under its own weight and initial ground stress. The convergence displacement along the shield axis increases with the distance from the excavation face. In the section where the convergence displacement is still less than the excavation gap, there is still a gap between the surrounding rock and the shield and no contact occurs, and the shield contact stress is zero. When the convergence displacement reaches the excavation gap, the surrounding rock begins to contact the outer surface of the shield and applies radial contact stress to the shield.

[0056] like Figure 14 As shown, within each creep time step, the distribution of the time-converged displacement of the surrounding rock along the shield axis is first obtained by numerical calculation. Next, find the intersection point of the distribution curve and the horizontal line of the excavation gap, and take the axial coordinate of the intersection point as the contact start coordinate. :when Less than hour Less than The surrounding rock and the shield did not make contact; when Greater than or equal to hour Greater than or equal to The surrounding rock comes into contact with the shield. Based on this, take... The section from the end L of the shield to the contact section is its length. As the downtime continued, The entire structure is raised, and the intersection point moves towards the front of the shield. As it decreases, the contact segment extends forward along the axial direction.

[0057] The integral of the shield contact stress along the shield surface therefore occurs only on the contact segment, and the lower limit of integration changes from 0 to... The section of the surrounding rock in the front part of the shield that has not yet converged to contact is not included in the integration, and its contact stress is originally zero. If it is still integrated according to the full length of the shield, the small contact stress generated on the grid of this section due to the tolerance of the contact algorithm will be included, so that the frictional resistance is systematically amplified with the length of the shield.

[0058] At the start of the current creep time step calculation, the shield contact stress obtained on the contact segment of the previous creep time step is used. A radial compressive stress boundary condition is applied to the surrounding rock surface of the contact section. This radial compressive stress raises the minimum principal stress of the surface element of the surrounding rock in the contact section from near zero to compressive stress, with its stress deviator being the second invariant. Consequently, the yield function of the viscoplastic activation criterion decreases. The value of decreases accordingly; when When the strain term changes from non-negative to negative, the viscoplastic deviatoric strain term of the improved nonlinear village hill mass is no longer activated in this element. The element only undergoes viscoelastic deviatoric strain updates, and its aging convergence rate is significantly slowed down. In other words, once the shield contacts the surrounding rock, its contact stress becomes the boundary condition constraining the subsequent aging deformation of that section of the surrounding rock; and the subsequent convergence displacement of that section of the surrounding rock determines the contact initiation coordinates and contact stress distribution of the next time step. This process continues step by step, with the contact segment range and contact stress distribution mutually constraining and simultaneously determined.

[0059] In the calculation, the creep time step is converted into actual time based on the construction tunneling speed. The creep time is advanced segment by segment and step by step according to the construction progress, ensuring that the creep time step corresponds to the actual time of tunneling and downtime periods. Assuming a construction tunneling speed of 12 m / day and an excavation advance of 1 m per step, the actual time required to complete one excavation step is 120 minutes. Taking each creep time step as 1 minute, each excavation step corresponds to 120 creep time steps. During downtime, the number of creep time steps is converted to the corresponding number of downtime periods for further iteration. During this period, the excavation face no longer advances, but the surrounding rock continues to undergo time-dependent deformation, and the contact start coordinate continues to move forward. The total number of iterations in the entire calculation process is determined by dividing the sum of the tunneling and downtime periods by the step size of the creep time step, and is a finite value.

[0060] In this embodiment, the selection of the creep time step satisfies the following conditions. First, since the contact stress is taken from the previous creep time step, the time step should be significantly smaller than the characteristic time of the viscoelastic response of the surrounding rock in the contact section. This characteristic time can be determined by the ratio of the Kelvin viscosity coefficient to the shear modulus. and The ratio is estimated; in engineering, the time step is usually no more than 1 / 10 of the characteristic time. Secondly, for the selected time step, it is halved and the same downtime period is recalculated. If the relative deviations of the contact start coordinates, total frictional resistance, and time-related jamming risk index do not exceed the given tolerance, for example, 1%, then the time step is considered sufficiently small. In this embodiment, the creep time step is 1 minute per step. After the above halving check, the relative deviations of the three quantities are all less than 1%. When the shield length, surrounding rock viscosity coefficient, or contact stiffness changes, the time step should be re-determined according to the above two conditions.

[0061] Step S6: Determine the total frictional resistance of the shield by integrating the shield contact stress along the shield surface on the contact section; determine the geometric margin by the contact start coordinate and shield length; determine the mechanical margin by the TBM rated thrust, total frictional resistance and thrust required for normal continuous tunneling; take the smaller of the geometric margin and the mechanical margin to determine the time-related jamming risk index and determine the jamming risk.

[0062] The total frictional resistance of the shield is obtained by integrating the contact stress along the shield surface on the contact section and adding it to the frictional resistance generated by the shield's own weight: ; In the formula For the shield radius, For shield length, For the initial contact coordinates, The coefficient of friction between the shield and the surrounding rock. axial coordinates Shield contact stress at the location, For the weight of the shield itself.

[0063] Whether the shield can pass through a certain tunnel section depends on whether two distinct constraints are simultaneously satisfied: First, the contact section must not extend to the full length of the shield; otherwise, the shield will be geometrically locked by the surrounding rock along its entire length. Second, the sum of the total frictional resistance of the shield and the thrust required for normal tunneling must not exceed the TBM's rated thrust; otherwise, the shield will be mechanically immobile. If either constraint is breached, the machine will jam, and the margin of one constraint cannot compensate for the deficiency of the other: when the contact section extends to the full length of the shield, even with ample thrust reserves, the newly exposed surrounding rock in front of the shield will immediately contact it, preventing the shield from moving forward; when the thrust reserves are exhausted, even if the front section of the shield has not yet made contact, the shield will still be immobile. Therefore, the geometric margin is defined by normalizing each constraint using its own allowable value. With mechanical margin : ; ; In the formula For the initial contact coordinates, For shield length, The thrust required for normal continuous tunneling The rated thrust of the TBM. Both margins are dimensionless quantities. A value of 1 indicates that the constraint has not been consumed at all, a value of 0 indicates that the constraint has just been exhausted, and a negative value indicates that it has been exceeded. Since the two constraints are hard constraints that do not compensate for each other in terms of value, their fusion cannot be done by weighted sum, as a weighted sum would cause the high margin of one constraint to mask the exhaustion of the other; nor can it be done by product, as a product would give too low a value when both constraints have moderate margins. The smaller of the two, i.e., the constraint closer to exhaustion, dominates the whole, which is consistent with the physical mechanism of machine jamming. Based on this, the time-sensitive machine jamming risk index is defined. : ; Dimensionless: when neither constraint has been consumed. Set to 0; when any constraint is exhausted. Take 1; A value greater than 1 indicates that a freeze has occurred. Therefore, the calculation is based on three mutually exclusive conditions: when... When the value is greater than or equal to 1, one of the two constraints has been broken, and a system freeze is detected; when Greater than or equal to the warning threshold When the value is less than 1, although the two constraints have not yet been broken, the margin is insufficient, which is judged as a high risk of card malfunction and a card malfunction warning is output; when Less than At that time, the risk of machine malfunction was at an acceptable level.

[0064] Its rate of change with creep time varies continuously with creep time The rate at which the tighter of the two constraints is consumed is given. When Less than and When the value is greater than zero, the remaining safe downtime is obtained by linear extrapolation at the current rate: ; The physical meaning is: if the machine stops from the current moment, Reaching the warning threshold The time that was previously sustainable; due to Less than 1, Inside The system is not yet at the critical point of freezing, so the duration is reduced by 1. Safety margin. Based on this, determine the shutdown window: planned shutdowns such as step changeovers and maintenance should be scheduled within [a certain timeframe]. Complete within; when If the downtime is shorter than the required downtime, the shield excavation gap should be increased before shutdown to raise the geometric margin and extend the operation period. .when When the value is less than or equal to zero, the aging convergence of the surrounding rock tends to stabilize or the contact section no longer expands, and the risk of jamming no longer increases. In this case, no calculation is performed. . From two adjacent creep time steps The difference quotient is obtained. To suppress the numerical jitter introduced by the contact algorithm, the average of the difference quotients of the most recent time steps is taken. In this embodiment, the time step is an integer not less than 3.

[0065] In one embodiment of the present invention, the above-mentioned linear extrapolation is... The improved nonlinear strain of the hill mass exhibits approximately linear growth within the extrapolated interval, a premise that holds true during both the decaying and steady-state creep stages. When the surrounding rock in the contact section enters the accelerated creep stage, the viscoplastic strain increases exponentially. As time increases, it exhibits an upward convex shape; at this point, the linear extrapolation obtained using the current rate of change... If the value is greater than the actual value, an unsafe shutdown window will be provided. Therefore, in The second difference is divided into positive, i.e. If the value is still increasing, substitute the maximum value of the difference quotient in the most recent few time steps into the above formula for calculation. .

[0066] The following is a complete collaborative work process based on the values ​​used in this embodiment: after the excavation face passes through a soft rock tunnel section, it enters the planned shutdown phase, and at the beginning of the shutdown, it contacts the initial coordinates. Located in the middle of the shield, that is The frictional resistance is 0.50L, the contact section length is approximately half the shield length, and the total frictional resistance is obtained by integrating the contact stress on the contact section. It is approximately 0.50 of the TBM's rated thrust. At this point, the geometric margin... The mechanical margin is 0.50. It is 0.37. The value was 0.63, below the threshold of 0.85. As the downtime increased, the yield function value of the viscoplastic activation criterion on the surrounding rock surface above the contact section decreased to a negative value due to contact stress constraints, and the time-dependent convergence rate slowed down. Meanwhile, the uncontacted section ahead of the contact section continued to develop according to free convergence, and its convergence displacement reached the excavation gap after 4 hours. The contact initiation coordinate shifted forward by approximately one-quarter of the shield length to 0.25L, and the length of the contact section increased accordingly. It rises to 0.63 of the rated thrust. At this point... It is 0.25. It is 0.24, take the smaller one. , It is 0.76. Based on the difference quotient between adjacent time steps, Approximately 0.033 per hour, remaining safe downtime (0.85) 0.76) / 0.033 is approximately 2.7 hours. Therefore, this shutdown should be completed within 2.7 hours, or the widening gap should be increased from 50mm to 70mm before shutdown; the latter shifts the contact initiation coordinate to 0.45L under the same convergence displacement distribution and shortens the contact section. It rose to 0.45. Reduced to 0.56 of rated thrust. If it rises to 0.31, the smaller value will still be used. , It dropped to 0.69. Extended to approximately 4.8 hours.

[0067] Compared with the approach of generating contact stress using elastoplastic constitutive model and integrating along the entire length of the shield to obtain frictional resistance, the contact stress obtained in this embodiment is a quantity that evolves over time, and its integration range changes with the expansion of the contact section. The frictional resistance and the surrounding rock convergence amount converge to a level that matches the actual contact state in the solution of time step by time.

[0068] To examine the applicability of the constitutive model to rocks of varying hardness, triaxial creep test data were used for five types of rocks: carbonaceous slate, mudstone, sandstone, argillaceous siltstone, and red mudstone. Rheological parameters were identified, and numerical reproductions were performed using the Deemo constitutive model. The comparisons between the calculated and experimental values ​​for mudstone, sandstone, argillaceous siltstone, and red mudstone are shown below. Figure 15 , Figure 16 , Figure 17and Figure 18 As shown. At the yield stress level, the determination coefficients of all the above-mentioned rocks are not less than 0.95. Taking mudstone as an example, in the unyielding grade with a confining pressure of 2 MPa and an axial stress of 4 MPa to 12 MPa, Take a range of 0.385 GPa to 1.03 GPa. Take a range of 0.167 GPa to 0.448 GPa. For stresses ranging from 0.968 GPa to 3 GPa, the coefficient of determination for each stress level ranged from 0.965 to 0.98; the coefficient of determination for stresses increasing to 14 MPa and reaching the yield level was 0.956. The results indicate that the constitutive model is applicable to rocks of different types and stress states.

[0069] In one embodiment of the present invention, in addition to the FLAC3D platform, other three-dimensional explicit finite difference platforms can also be used, such as those that solve the equations of motion with explicit time-stepping, where constitutive modules and solvers are independent and have open custom constitutive dynamic link library interfaces. In this case, the viscoelastic-plastic creep constitutive model built into the platform is used as a blueprint, and the aforementioned nine core member functions are implemented according to its constitutive interface specification and compiled into a dynamic link library, which is then called by the platform's plugin loading and constitutive assignment instructions. As another implementation, in addition to using the Levenberg-Mquardt algorithm, the rheological parameter identification can also use a trust region reflection algorithm, which belongs to the damped nonlinear least squares algorithm. This algorithm uses the trust region radius instead of the damping step size to constrain the iteration step size, and the parameter values ​​and determination coefficients obtained from the creep curves of the same group under staged loading are comparable to those of the Levenberg-Mquardt algorithm.

[0070] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; only preferred embodiments of the present invention are illustrated. The descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. As long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0071] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A risk prediction method for shield-type TBM haulage machines considering the time-dependent deformation of surrounding rock, characterized in that, Includes the following steps: Step S1: Construct a nonlinear rheological constitutive model consisting of an improved nonlinear village mountain and a Burgers model connected in series. The Burgers model consists of a Maxwell body and a Kelvin body connected in series. The improved nonlinear village mountain consists of a plastic switch and a series branch connected in parallel. The series branch consists of a Kelvin body and an exponentially unsteady Newtonian body whose viscosity decays with time connected in series. Step S2: Identify the rheological parameters of the constitutive model based on the rock graded loading creep test curve; Step S3: The three-dimensional creep equation of the constitutive model is derived into an incremental finite difference scheme, and the yield criterion on which the viscoplastic activation criterion and post-yield stress correction are based is determined. Based on this, a custom constitutive model is obtained through secondary development on a three-dimensional explicit finite difference platform. Step S4: In the shield-type TBM dynamic excavation numerical model, call the custom constitutive model to solve the axial distribution of the surrounding rock age convergence displacement and the shield contact stress step by step in the creep time step; Step S5: In each creep time step, take the axial coordinate at the point where the time-effect convergence displacement of the surrounding rock is equal to the shield excavation gap as the contact start coordinate, and take the section from the contact start coordinate to the tail end of the shield as the contact section; apply the shield contact stress on the contact section of the previous creep time step as the radial compressive stress boundary condition to the surface of the surrounding rock of the contact section. Step S6: Determine the total frictional resistance of the shield by integrating the shield contact stress along the shield surface on the contact section; determine the geometric margin by the contact start coordinate and the shield length; determine the mechanical margin by the TBM rated thrust, the total frictional resistance and the thrust required for normal continuous tunneling; take the smaller of the geometric margin and the mechanical margin to determine the time-related jamming risk index and determine the jamming risk.

2. The method according to claim 1, characterized in that, The one-dimensional creep constitutive equation of the constitutive model is piecewise: When the initial stress is less than the yield strength: ; When the initial stress is greater than or equal to the yield strength, an improved nonlinear hill-mountain term is superimposed: ; in For the total strain of the model, For stress, For initial stress, For yield strength, Creep time, The attenuation coefficient is... and These represent the elastic modulus and viscosity coefficient of the Maxwell body in the Burgers model, respectively. and These represent the elastic modulus and viscosity coefficient of the Kelvin matrix in the Burgers model, respectively. and These represent the elastic modulus and viscosity coefficient of the Kelvin matrix in the improved nonlinear village hill, respectively. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. For switching functions, when Less than or equal to When taking 0, Greater than Time to take .

3. The method according to claim 1, characterized in that, The three-dimensional creep equation of the constitutive model is obtained by decomposing the stress tensor into a spherical tensor and a deviatoric tensor. The spherical strain is completed instantaneously and does not change with time, while the creep is caused only by the deviatoric stress tensor. The three-dimensional deviatoric strain components of the improved nonlinear village hill for: ; in For the deviatoric stress tensor, For yield deviatoric stress, The shear modulus of the Kelvin form in the improved nonlinear village hill is given. The viscosity coefficient of the Kelvin form in the improved nonlinear village hill is given. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. The attenuation coefficient is... Creep time; The yield function of the viscoplastic activation criterion is: , As the second invariant of stress deviator, For yield strength, when The viscoplastic deviatoric strain term of the improved nonlinear village hill is activated when it is greater than or equal to 0. When the value is less than 0, only viscoelastic deviatoric strain is updated.

4. The method according to claim 1, characterized in that, In step S2, the Boltzmann superposition principle is first used to convert the graded loading creep curves of the rock into separate loading creep curves under each stress level. Then, the damped Levenberg-Marquardt algorithm is used to identify the rheological parameters. The iterative process is expressed as follows: ; in Let be the parameter vector for the s-th iteration. Let Hessian matrix be the objective function. The first gradient of the objective function. For the damping step size, It is an identity matrix.

5. The method according to claim 1, characterized in that, In step S3, the incremental finite difference scheme decomposes the total deviatoric strain increment into the sum of four deviatoric strain increments: instantaneous elastic body, Newtonian body, viscoelastic body and improved nonlinear body, and approximates the deviatoric stress and deviatoric strain within the time step with the mean of the new and old values ​​at the time step. The deviatoric stress update formula for the entire constitutive model is obtained when the deviatoric stress has not reached the yield deviatoric stress: ; When the deviatoric stress reaches the yield deviatoric stress: ; in This is the deviatoric stress tensor, where the superscripts N and O represent the new and old values ​​at the time step, respectively. This represents the total partial strain increment. The deviatoric strain increment of the improved nonlinear village hill. This represents the deviatoric strain aging value for a viscoelastic body. and These are intermediate coefficients determined by the shear modulus, viscosity coefficient, and time step in the Kelvin model of the Burgers model. and The coefficients are determined by each elastic shear modulus, viscosity coefficient, time step, and the intermediate coefficients.

6. The method according to claim 1, characterized in that, The yield criterion upon which the post-yield stress correction is based is the Mohr-Coulomb criterion with tensile intercept. The Mohr-Coulomb criterion includes a shear instability criterion and a tensile instability criterion. The shear instability criterion is... The tensile instability criterion is The boundary function between the shear plastic region and the tensile plastic region is: The uncorrelated shear potential function is The tension potential function is ,in and These are the maximum principal stress and the minimum principal stress, respectively. For cohesion, It is the internal friction angle. For shear expansion angle, For tensile strength, The coefficient is determined by the internal friction angle. The coefficient is determined by the shear dilatation angle. and These are used to determine the intermediate values ​​of the slope and intercept of the boundary function, respectively. The post-yield stress correction first calculates the test stress using viscoelasticity and then determines the principal stresses of the test stress. Next, it uses the shear instability criterion and the tensile instability criterion to determine whether the test stress exceeds the yield surface. Less than or equal to 0 and When the test stress is greater than or equal to 0, it is determined that the test stress has not exceeded the yield surface, and the test stress is taken as the final stress. When the test stress exceeds the yield surface and When the stress is less than 0, the test stress is determined to be located in the shear-plastic region. The test stress is then corrected back to the yield surface according to the shear potential function and the flow rule. The shear-plastic multiplier used for this correction... for: ; in and These are the maximum and minimum test principal stresses, respectively. and These represent the shear modulus and viscosity coefficient of the Kelvin matrix in the improved nonlinear village hill, respectively. The initial value of the viscosity coefficient of the unsteady Newtonian body in the improved nonlinear village hill is given. The attenuation coefficient is... The step size of the creep time step; When the test stress exceeds the yield surface and When the value is greater than 0, it is determined that the test stress is located in the tensile plastic region, and the test stress is corrected back to the yield surface according to the tensile potential function and the flow law.

7. The method according to claim 1, characterized in that, In step S6, the total frictional resistance experienced by the shield Calculate using the following formula: ; in For the shield radius, For shield length, The contact start coordinates are... The coefficient of friction between the shield and the surrounding rock. Coordinates along the shield axis Shield contact stress at the location, For the weight of the shield; The geometric margin is The mechanical margin is The time-sensitive card machine risk index is ; in The thrust required for normal continuous tunneling Rated thrust of the TBM The value is 1 when the shield has not yet made contact with the surrounding rock, and 0 when the contact section extends to the full length of the shield, i.e., the shield is completely covered by the surrounding rock; when A machine freeze is determined when the value is greater than or equal to 1; when Greater than or equal to the warning threshold And when it is less than 1, it is judged as a high risk of card malfunction and a card malfunction warning is issued; when Less than and Rate of change of creep time When greater than zero, press Calculate the remaining safe downtime, and determine the shutdown window or increase the shield excavation gap according to the remaining safe downtime; when When the value is less than or equal to zero, the risk of machine malfunction is determined to no longer increase.

8. The method according to claim 1, characterized in that, The constitutive secondary development in step S3 includes: declaring all parameters of the constitutive model according to the constitutive interface specification of the three-dimensional explicit finite difference platform, implementing member functions for initializing large deformation states and member functions for updating stress tensors and strain tensors step by step, compiling to generate a dynamic link library, loading the dynamic link library through the platform's plug-in configuration instructions, and assigning the custom constitutive model to the computational domain using constitutive assignment instructions.

9. The method according to claim 1, characterized in that, In step S4, the creep time step is converted into actual time according to the construction tunneling speed, and the creep time is advanced step by step according to the construction progress. The number of iterations of each creep time step is determined by dividing the sum of the tunneling period and the shutdown period by the step size of the creep time step. The step size of the creep time step is not greater than 1 / 10 of the ratio of the viscosity coefficient of the Kelvin body in the Burgers model to the shear modulus of the Kelvin body in the Burgers model. After the step size is halved, the relative deviations of the contact start coordinate, the total frictional resistance of the shield and the time-related jamming risk index do not exceed the preset tolerance.

10. The method according to claim 7, characterized in that, when When the second difference of the creep time is greater than zero, take the most recent creep time steps. Substituting the maximum value into the formula for calculating the remaining safe downtime, we obtain the remaining safe downtime.