Reliability robust design method for multiple failure modes of ultra-deep well hoisting container

Abstract

A reliability robust design method for multiple failure modes of an ultra-deep well hoisting container, including: defining randomness of a structural parameter, a material property, and a dynamic load of a hoisting container, and solving a random response of a structural failure for a random parameter using a design of experiment method; establishing reliability performance functions of each failure modes in accordance with failure criterion of the hoisting container; establishing a joint probability model of correlated failures using a copula theory in consideration of probability correlation between the failure modes; establishing, a system reliability model with failure correlation of the hoister container; establishing a sensitivity model concerning each random parameter for system reliability of the hoisting container; and establishing, in conjunction with an optimization design model, a reliability robust optimization design model for the hoisting container using a joint failure probability and parameter sensitivity as constraints.

Claims

1. A reliability robust design method for multiple failure modes of an ultra-deep well hoisting container, comprising the following steps: step 1: determining means and variances of basic parameters including a dimension parameter, a material property parameter, and a load of the ultra-deep well hoisting container, determining a distribution type of each parameter, and establishing a finite element model of the ultra-deep well hoisting container; step 2: obtaining a random response sample of a structural failure for each set of driving parameters according to the means and the variances of the basic parameters of the ultra-deep well hoisting container determined in step 1 in conjunction with a Latin hypercube sampling design of experiment method; step 3: using input and output samples in step 2 by using a Kriging method, to obtain a mapping relationship between a failure response and a structural performance parameter of the ultra-deep well hoisting container, and establishing reliability functions in the failure modes in accordance with a design criteria of failures of the hoisting container; step 4: respectively solving, according to probability information of the basic parameters, failure probabilities of the failure modes by using a moment-based saddlepoint approximation method; step 5: establishing a joint failure distribution with probability correlation of the failure modes by using a Clayton copula function, establishing a system reliability model in a joint probability failure, and solving a system failure probability of the ultra-deep well hoisting container; step 6: establishing a sensitivity model concerning a random parameter for system reliability of the ultra-deep well hoisting container by using a partial derivative; and step 7: establishing the reliability robust design model of the ultra-deep well hoisting container by using the sensitivity model concerning the random parameter for the system reliability and the system failure probability of the ultra-deep well hoisting container obtained in step 5 and step 6 used as constraints, wherein the step 5 further comprising: performing a random sampling by using a uniform sampling method in accordance with a distribution type of a random variable of the ultra-deep well hoisting container, to obtain a discrete sample value of each random variable; using the discrete sample value in a reliability function of the reliability functions established in step 3, to obtain a corresponding function sample value; calculating a rank correlation coefficient between two failure modes, using the rank correlation coefficient in a Clayton copula function model, to calculate a parameter to be determined of the Clayton copula, and establishing a joint probability model for describing probability correlation; calculating a joint failure probability with correlation of multiple failure modes of the ultra-deep well hoisting container by using the joint probability model; and calculating the system failure probability of the ultra-deep well hoisting container.

2. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 1 further comprising: determining distribution types, means, and variances of a structural dimension and a material property of the ultra-deep well hoisting container; determining the load of the ultra-deep well hoisting container, to determine distribution types, means, and variances of loads, including a static load, a bending moment, and a torque, borne by the ultra-deep well hoisting container in each case; and establishing a finite element analysis model of the ultra-deep well hoisting container based on the foregoing information.

3. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 2 further comprising: forming a process file of modeling by parameterized modeling of the ultra-deep well hoisting container; forming a process file of finite element analysis by finite element analysis of the ultra-deep well hoisting element, wherein a structural parameter of the ultra-deep well hoisting container comprises a dimension of the ultra-deep well hoisting container and a dimension of a chassis; and a material performance parameter comprises an elastic modulus, a Poisson's ratio, and density; and driving, by using a Latin hypercube sampling design of experiment method, the parameters of the ultra-deep well hoisting container to perform random finite element analysis, to obtain a random response sample with a random input.

4. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 3 further comprising: using the input and output samples obtained in step 2 by using the Kriging method to establish an explicit function relationship between a random response and a random parameter; and establishing the reliability functions in the failure modes in accordance with the design criteria of the failure modes, wherein the ultra-deep well hoisting container is a large-scale welded structural component, and when strength reliability of the ultra-deep well hoisting container is investigated, fracture mechanics analysis is used, and fracture resistance performance of the ultra-deep well hoisting container is used as a criterion for strength design.

5. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 4 further comprising: calculating first three order moments, including a mean, a variance, and skewness, of each function by using a random perturbation technology, and solving the failure probabilities of the failure modes by using a saddlepoint approximation method based on the first three order moments.

6. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 6 further comprising: on the basis of establishing a system failure probability of the ultra-deep well hoisting container, performing derivation on a mean, a standard deviation, and a skewness of a random variable by using a matrix calculus according to the partial derivative, to establish a parameter sensitivity model concerning the mean, the standard deviation, and the skewness of the random variable for the system reliability of the ultra-deep well hoisting container.

7. The reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to claim 1, wherein the step 7 further comprising: introducing a system reliability and a parameter reliability sensitivity model obtained and are based on a copula function into an optimization design model as constraints, to establish the reliability robust design model of the ultra-deep well hoisting container.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a flowchart of implementation of a reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to the present invention;

(2) FIG. 2 is a schematic structural diagram of a hoisting container;

(3) FIG. 3 is a probability density diagram of a Clayton copula function; and

(4) FIG. 4 is a scatter diagram of a Clayton copula function.

(5) 1: upper tray; 2: high-strength bolt set; 3: middle tray; 4: external column; 5: tank wall; 6: internal column; 7: lower tray.

DETAILED DESCRIPTION OF THE EMBODIMENTS

(6) The present invention is further described below with reference to the accompanying drawings and embodiments.

(7) As shown in FIG. 1, a reliability robust design method for multiple failure modes of an ultra-deep well hoisting container according to the present invention includes the following steps:

(8) Step 1: Obtain means and variances of parameters, such as a structural dimension, a material property, and a dynamic load, of a hoisting container according to an original design drawing of the hoisting container or by on-site surveying and mapping.

(9) Step 2: Establish a three-dimensional parameterized model of the hoisting container according to the structural parameter of the hoisting container, and perform finite element static analysis on the established virtual prototype model.

(10) Step 3: Establish, by using a design of experiment method, random sampling matrixes of basic parameters according to the means and the variances of the basic parameters of the hoisting container determined in step 1 in conjunction with a Latin hypercube sampling method.

(11) Step 4: Obtain a stress strength factor and a strain random response sample of the hoisting container by means of model reconstruction and finite element reanalysis in conjunction with an experiment design matrix.

(12) Step 5: Fit the experiment design matrix and the random response sample by using a Kriging method, so as to establish a mapping relationship between a random response and a random parameter of the hoisting container.

(13) Step 6: Respectively establish reliability performance functions in a fracture failure and in a stiffness failure in accordance with a damage tolerance criterion of crack extension and a stiffness design criterion of the hoisting container; and calculate the first three order moments of the random parameter according to a mean and a variance of the random parameter, so as to solve a mean, a variance, and three order moment of each performance function according to the established function, and respectively calculate failure probabilities in two failure modes by using a saddlepoint approximation method based on the first three order moments.

(14) Step 7: Obtain a correlation coefficient between the two failure modes by using a statistics method, establish a joint probability model thereof by using a Clayton copula, and solve a system failure probability with failure correlation in conjunction with a system reliability method.

(15) Step 8: Establish a parameter sensitivity model concerning a mean, a standard deviation, and a skewness of a random variable for system reliability by using a matrix differential technology.

(16) Step 9: Introduce the system reliability and the parameter sensitivity model into an optimization design model, to establish the reliability robust design model of the hoisting container.

Embodiments

(17) To understand the features and engineering applicability of the present invention more sufficiently, in the present invention, for an ultra-deep well hoisting container structure to be constructed shown in FIG. 2, system reliability analysis and structural design with failure correlation are performed.

(18) The hoisting container bears a vertical load and a bending-torsional coupling effect. An experiment design sample matrix of a random variable of the hoisting container is established according to a structural dimension and a dynamic load of the hoisting container, and a stress strength factor and a strain random response of the hoisting container are calculated by finite element analysis, to obtain a response sample matrix. A fitting function between a random response and an experiment design sample matrix is established by using a Kriging method, so as to establish explicit functions in two failure modes, namely, a fracture strength failure and a stiffness failure, in accordance with a damage tolerance criterion and a stiffness criterion. Table 1 provides probability information of random variables of the hoisting container in the present embodiment. L.sub.1 is a length of the hoisting container, L.sub.2 is a height of the hoisting container, L.sub.3 is a width of the hoisting container, L.sub.4 is a length of a lower tray, and L.sub.5 is a width of the lower tray.

(19) TABLE-US-00001 TABLE 1 Probability statistical characteristics of random variables of the hoisting container Variable Mean Standard deviation Distribution type L.sub.1 (mm) 4000 200 Normal L.sub.2 (mm) 3170 158.5 Normal L.sub.3 (mm) 2050 102.5 Normal L.sub.4 (mm) 4000 200 Normal L.sub.5 (mm) 1800 90 Normal

(20) In this embodiment, a failure probability in a fracture strength failure mode, namely, Pf.sub.1=0.004698, and a stiffness failure probability, namely, Pf.sub.2=0.007344, are obtained by using a solution method for a failure probability provided in the present invention. The n random samples of random variables of the hoisting container are generated according to distribution types of the random variables, and are respectively substituted into functions in the two failure modes, to calculate n response values. A correlation coefficient between random response samples in the two failure modes is obtained by MATLAB calculation, and a parameter to be determined of the Clayton copula function is estimated by using the correlation coefficient, so as to establish a joint probability model with failure correlation of the hoisting container. The obtained failure probabilities Pf.sub.1 and Pf.sub.2 are substituted into a system reliability analysis model.

(21) $P_{f1}+\underset{i=2}{\overset{m}{.Math.}}\max \left(P_{\mathrm{fi}}-\underset{j=1}{\overset{i-1}{.Math.}}{P}_{\mathrm{fij}},0\right)P_{\mathrm{fi}}\underset{i=1}{\overset{m}{.Math.}}{P}_{\mathrm{fi}}-\underset{i=2}{\overset{m}{.Math.}}\max \left(P_{\mathrm{fij}}\right)$

(22) In the formula, m=2 indicates a number of failure modes of the hoisting container, P.sub.f1 indicates a maximum failure probability in the failure modes of the hoisting container, P.sub.fi indicates a failure probability of the i.sup.th failure mode, Py.sub.fij indicates a joint failure probability of the i.sup.th and the j.sup.th failure modes, and P.sub.fs indicates a system failure probability with failure correlation of the hoisting container.

(23) The system failure probability, namely, P.sub.fs=0.01319, of the hoisting container with correlation between the fracture strength failure and the stiffness failure of the hoisting container is calculated based on a Clayton copula-based joint probability model. The system failure probability, namely, P.sub.fsm=0.01142, of the hoisting container is calculated by using a simulation method for 1000 times.

(24) Parameter sensitivity concerning a mean, a standard deviation, and a skewness of a random variable of the system failure probability is calculated below:
p.sub.f/.sub.L1=2.72910.sup.1, p.sub.f/.sub.L1=8.41410.sup.2, p.sub.f/.sub.L1=1.16510.sup.3
p.sub.f/.sub.L2=5.25110.sup.2, p.sub.f/.sub.L2=2.66810.sup.3, p.sub.f/.sub.L2=5.12510.sup.3
p.sub.f/.sub.L3=5.15610.sup.2, p.sub.f/.sub.L3=3.41010.sup.2, p.sub.f/.sub.L3=3.84710.sup.2
p.sub.f/.sub.L4=7.85910.sup.1, p.sub.f/.sub.L4=5.87510.sup.1, p.sub.f/.sub.L4=2.65110.sup.2
p.sub.f/.sub.L5=3.61410.sup.1, p.sub.f/.sub.L5=3.01810.sup.1, p.sub.f/.sub.L5=6.33110.sup.3

(25) With a total volume of the hoisting container as an optimization object, and system reliability and parameter sensitivity of a structural system as constraint conditions, an optimized structural parameter combination can be obtained by using a non-linear optimization method as follows: L.sub.1=3817.413 mm, L.sub.2=3119.55 mm, L.sub.3=2112.93 mm, L.sub.4=3817.41 mm. L.sub.5=1850.01 mm

(26) A total volume of the hoisting container before optimization is 15.99 m.sup.3, and after reliability robust optimization is used, under the condition of satisfying reliability constraints in an optimization model, a total volume of a hoisting container structure is 15.16 m.sup.3.

(27) A total volume of an optimized structure obtained by using a conventional optimization design method is 15.37 m.sup.3.

(28) It can be seen by comparing the foregoing results that the reliability robust design method provided in the present invention can produce a better optimization result.

(29) In conclusion, the present invention provides a system reliability robust design method for an ultra-deep well hoisting container in consideration of failure correlation. Firstly, a three-dimensional parameterized model of a hoisting container is established according to a structural dimension of the hoisting container; secondly, experiment design sample matrixes of random variables are established according to probability properties of random variables of the hoisting container, and fracture strength and stiffness response of the hoisting container in the sample matrixes are solved by using a finite element method; thirdly, an explicit function between a random response and a random variable matrix is established by using a Kriging method, and explicit functions in two failure modes are respectively established in accordance with a damage tolerance criterion and a stiffness design criterion; then, failure probabilities in the two failure modes are calculated by using a saddlepoint approximation method based on the first three order moments, a joint failure probability model between the two failure modes is constructed by using a Clayton copula function, and system reliability in a joint failure is solved by using a system reliability method; and finally, parameter sensitivity concerning the random variables for the system reliability is calculated, and an optimal parameter combination is obtained by using a reliability robust design model.

(30) Those that are not described in detail in the present invention are known to a person skilled in the art.