基于k-fold交叉验证的代理模型序列采样方法

在代理模型序列采样框架下,针对现有研究中的不足之处,通过引入k-fold交叉验证计算样本的预测误差,并结合泰森多边形法和最大距离最小化准则,发展了一种适用于任意代理模型的k-fold CV-Voronoi自适应序列采样方法。相较于传统序列采样方法,本文方法具有计算简单和自适应性强等显著优势。通过数值算例和工程算例对比分析发现所提序列采样方法具有较高的近似精度和计算效率,此外,进一步讨论了k-fold交叉验证中k的不同取值对于代理模型精度的影响,总结出k的最优取值范围以供参考。     

基于主动学习Kriging模型与序列重要抽样的随机-区间混合可靠性分析

针对随机-区间混合可靠性分析中复杂功能函数的高非线性和多设计点问题,本文提出了一种结合主动学习Kriging模型与序列重要抽样方法的混合可靠性分析方法.在序列重要采样方法中采用高斯混合分布作为提议分布进行逐级采样,逐步逼近最优重要抽样函数的采样样本;结合序列重要抽样方法的特点,提出了主动学习Kriging模型的两步学习...     

波纹板结构动力学建模及其自由振动特性分析

波纹板结构是高速列车中较为常见的型材结构,因其具有轻质、高强度和高稳定性等一系列优点,所以广泛应用于车体结构轻量化设计。首先,基于一阶剪切变形理论和中厚壳(圆弧壳)理论,采用微分求积有限元法(DQFEM)建立一般边界条件下五自由度波纹板结构的通用动力学分析模型;紧接着对五自由度波纹板结构动力学模型进行收敛性验证,确定最佳微分节点数和惩罚因子的取值;然后,采用有限软件(ABAQUS)验证了建立模型的准确性;最后根据已建立模型,研究波纹板结构参数对其振动特性的影响。结果表明,波纹板结构厚度和固支边界条件会增加其固有频率,提高波纹板结构的稳定性,使得共振频率向高频移动;增加波纹板结构长度、参与耦合圆弧壳半径和参与耦合圆弧壳弧度会降低其固有频率,使得共振频率向低频移动。     

Upscaling and Simulation of Waterflooding in Heterogeneous Reservoirs Using Wavelet Transformations: Application to the SPE-10 Model

We have developed an accurate and highly efficient method for upscaling and simulation of immiscible displacements in three-dimensional (3D) heterogeneous reservoirs, which is an extension of the technique that we developed previously for 2D systems. The method utilizes wavelet transformations (WTs) to upscale the geological model of a reservoir, based on the spatial distribution of the single-phase permeabilities and the locations of the wells in the reservoir. It generates a non-uniform grid in which the resolved structure of the fine grid around the wells, as well as in the high-permeability sectors, are preserved, but the rest of the grid is upscaled. A robust uplayering procedure is used to reduce the number of the layers, and the WTs are used to upscale each layer areally. To demonstrate the method’s accuracy and efficiency, we have applied it to the geological model of a highly heterogeneous reservoir put forward in the tenth Society of Petroleum Engineers comparative solution project (the SPE-10 model), and carried out simulation of waterflooding in the upscaled model. Various upscaling scenarios were examined, and although some of them resulted in efficient simulations and accurate predictions, the results when non-uniform upscaling is used based on the WT technique are in excellent agreement with the solution of the same problem in the fine grid of the SPE-10 model. Most importantly, the speed-up factors that we obtain are several orders of magnitude. Hence, the method renders it unnecessary to use massively parallel computations for such problems.     

Constructal multi-scale structure of PCM-based heat sinks

This paper inquires the effectiveness of a PCM-based heat sink as a reliable solution to portable electronic devices. This sink is composed of a PCM with low thermal conductivity and fins to boost its conductivity. The optimization is subjected to fixed heat sink volume filled with PCM between vertical equidistant fins. New fins are installed in the unheated space existing in each enclosure which is not involved in thermal distribution from vertical fins to the PCM. Based on the same principle, new fins generations are augmented stepwise to the multi-scale structure. The steps of adding fins will continue up to the point that the objective function reaches its maximal value, i.e., maximizing the longest safe operation time without allowing the electronics to reach the critical temperature. The results indicate that in each length of the enclosure, the optimum volume fraction and the best fins distance values exist in which the heat sink performance becomes maximum, and adding more fins lowers the performance of the heat sink. Increasing the enclosure’s length by \(2^{n}\) does not change them. For an enclosure with constant length, the optimal number of steps for adding fins within the enclosure is a function of the fin thickness. The results indicate that increasing the thickness changes the optimal number of adding fins inside the enclosure (normally a decrease). As the fin thickness is lowered, there will be a higher effect by adding vertical fins in the enclosure. Numerical simulations cover the Rayleigh number range \(2\times 10^{5}\le \hbox {Ra}_{\mathrm{H}} \le 2.7\times 10^{8}\), where H is the heat sink height.     

Accuracy of the Sampling Moiré Method and its Application to Deflection Measurements of Large-Scale Structures

Measuring accurate displacement distributions for large-scale structures is an important issue and a very challenging task. Recently, a simple and accurate phase measurement technique called sampling moiré method [Exp Mech 50–4:501–508, (2010)] has been developed for small-displacement distribution measurements. In this method, the phase distribution of moiré fringes can be analyzed from a single grating image by simultaneously performing down-sampling image processing and intensity-interpolation to generate multiple phase-shifted moiré fringe images. In addition, the phase of the original grating can also be obtained from the phase of the moiré fringe by adding the phase of the sampling grating. In this study, the measurement accuracy of the sampling moiré method was analyzed through computer simulations and a displacement measurement experiment. Four factors of the sampling moiré method were investigated, including the sampling pitch, the order of the intensity-interpolation, random noise, and the form of grating. The results show that determining the optimal sampling pitch is an important factor for obtaining better results but it is not critical. In addition, a practical application of the sampling moiré method is presented that involves a deflection measurement on a 10-meter-long crane. The experimental results demonstrate that submillimeter deflections of the crane can be successfully detected.     

Domain decomposition method with hybrid approximations applied to solve problems of elasticity

To solve two-dimensional boundary-value problems of elasticity, two iteration algorithms of the domain decomposition method are proposed: parallel Neumann–Neumann and sequential Dirichlet–Neumann. They are based on the hybrid boundary–finite-element approximations. The algorithms are proved to converge. The optimal parameters are selected using the minimum-residual and steepest-descent methods. Some plane problems of elasticity are solved as examples, and stationary and nonstationary iteration algorithms in these examples are analyzed for efficiency Translated from Prikladnaya Mekhanika, Vol. 44, No. 11, pp. 18–29, November 2008.     

Comparison of an Adaptive Wavelet Method and Nonlinearly Filtered Pseudospectral Methods for Two-Dimensional Turbulence

An adaptive wavelet method for solving the two-dimensional Navier–Stokes equations is compared with nonlinear Fourier filtering and nonlinear wavelet filtering of the pseudospectral method at each time step. The methods are each applied to a highly nonlinear flow typical of two-dimensional turbulence, the merger of two positive vortices pushed together by a weaker negative vortex, and the results are compared with a reference classical pseudospectral method. Nonlinear Fourier filtering uses 1.7 times fewer active modes than the reference simulation at the time of merger (when the flow is most complicated) and retains the overall dynamics and structure of the flow. However, it induces spurious oscillations in the background. Nonlinear wavelet filtering simulation uses 9.2 times fewer modes than the reference simulation at the time of merger, and reduces the errors in the solution. The adaptive wavelet simulation replicates precisely the dynamics and spatial structure of the reference simulation while retaining the high compression rate of the nonlinear wavelet filtering simulation. In addition we observe that the number of active wavelet modes remains quasi-constant during the whole merging process, independent of the strength of the vorticity gradients. On the contrary, the number of active Fourier modes is multiplied by 5 when the vorticity gradients are strongest. The increased accuracy of the adaptive wavelet simulation is due to the security zone added around the active coefficients and to the compression of the nonlinear term of the Navier–Stokes equations in the wavelet basis. These results suggest that nonlinear Fourier filtering of a classical pseudospectral method cannot produce significant improvement, but that the adaptive wavelet method combines a consistently high compression rate with high accuracy. Received 22 April 1997 and accepted 11 August 1997     

On one property of positive-definite quadratic forms corresponding to finite posets

We consider the Bondarenko hypothesis about the structure of finite posets with positive-definite Tits quadratic form. We prove that the hypothesis holds for a fixed natural number k > 8 if it holds for the number k − 1. __________ Translated from Neliniini Kolyvannya, Vol. 8, No. 4, pp. 544–552, October–December, 2005.     

DAMAGE DETECTION BASED ON OPTIMIZED INCOMPLETE MODE SHAPE AND FREQUENCY

For the purpose of structural health monitoring, a damage detection method combined with optimum sensor placement is proposed in this paper. The back sequential sensor placement(BSSP) algorithm is introduced to optimize the sensor locations with the aim of maximizing the 2-norm of information matrix, since the EI method is not suitable for optimum sensor placement based on eigenvector sensitivity analysis. Structural damage detection is carried out based on the respective advantages of mode shape and frequency. The optimized incomplete mode shapes yielded from the optimal sensor locations are used to localize structural damage. After the potential damage elements have been preliminarily identified, an iteration scheme is adopted to estimate the damage extent of the potential damage elements based on the changes in the frequency. The effectiveness of this method is demonstrated using a numerical example of a 31-bar truss structure.     

Application of an Inverse Simulator of Single-Phase Flow Analysis for Leakage Pathway Estimation in a Multiphase System for Geologic $$\hbox {CO}_{2}$$ Storage

When \(\hbox {CO}_{2}\) is injected in a brine reservoir, brine or \(\hbox {CO}_{2}\) can be discharged into a permeable formation saturated with brine above the storage reservoir along a leakage pathway, if present. In most situations, the overlying formation can act as a single-phase aquifer with only brine leakage before \(\hbox {CO}_{2}\) leaks. This study examines the applicability of a developed inverse code for single-phase fluids to detect leakage pathway locations in view of the overlying formation using pressure anomalies induced by leaks. Before applying inverse analysis, forward modeling is performed using the TOUGH2 model to determine brine and \(\hbox {CO}_{2}\) leakage in a homogeneous conceptual model, and the simulated pressure profiles at monitoring wells are used as measurements in the inverse model. In the inverse code, an important consideration is that the vertical hydraulic conductivity and cross-sectional area of a leakage pathway that are inherent to a leakage term in the mass balance equation are integrated as a single parameter to estimate the leakage pathway locations. This method eliminates the impact of the uncertainty of the leakage pathway size on the accuracy of leakage pathway estimation. The inverse model examines the effect of the number of monitoring wells, monitoring periods and \(\hbox {CO}_{2}\) leakage into the overlying formation on the accuracy of leakage pathway estimation according to eleven application examples. The comparison between the results of the single-phase inverse code and iTOUGH2 code illustrates that the single-phase inverse model can be applicable to the leakage pathway estimation in a multiphase flow system.     

Error Assessment in Stereo-based Deformation Measurements

Increasing interest in the use of digital image correlation (DIC) for full-field surface shape and deformation measurements has led to an on-going need for both the development of theoretical formulae capable of providing quantitative confidence margins and controlled experiments for validation of the theoretical predictions. In the enclosed work, a series of stereo vision experiments are performed in a manner that provides sufficient information for direct comparison with theoretical predictions using formulae developed in Part I. Specifically, experiments are performed to obtain appropriate optimal estimates and the uncertainty margins for the image locations/displacements, 3-D locations/displacements and strains when using the method of subset-based digital image correlation for image matching. The uncertainty of locating the 3-D space points using subset-based pattern matching is estimated by using theoretical formulae developed in Part I and the experimentally defined confidence margins for image locations. Finally, the uncertainty in strains is predicted using formulae that involves both the variance and covariance of intermediate variables during the strain calculation process. Results from both theoretical predictions and the experimental work show the feasibility and accuracy of the predictive formulae for estimating the uncertainty in the stereo-based deformation measurements.     

斜拉桥合理成桥状态确定的序列二次规划法

本文首先介绍斜拉桥合理成桥状态的概念和现有的斜拉桥索力优化方法。然后基于序列二次规划(SQP)算法,提出了一种用于确定斜拉桥成桥合理状态的实用方法,序列二次规划法。该方法通过建立斜拉桥索力优化的非线性规划模型,以斜拉桥主梁和索塔的弯曲应变能为目标函数,以各斜拉索的索力为设计变量,结构的应力和索力为约束条件,并计人大跨度斜拉桥各种几何非线性因素的影响,采用强次可行序列二次规划法进行优化求解,确定斜拉桥成桥合理状态的索力。运用该方法和空间非线性有限元分析程序分析了某斜拉桥的合理成桥状态,计算结果表明该方法简单、有效。     

利用线性模型估计的传感器优化布置算法

提出了一种基于线性模型估计的传感器布置算法.首先根据线性模型估计理论,将待监测的目标模态振型视为线性模型的设计矩阵;然后利用奇异值分解的算法,将设计矩阵分解,根据分解的前几个左奇异向量来计算各个自由度对于目标模态振型的贡献;用迭代算法来求出最优的传感器布置方案;最后,用两个算例表明该方法的有效性.     

Actuation of the Kagome Double-Layer Grid. Part 1: Prediction of performance of the perfect structure

The Kagome Double-Layer Grid (KDLG) is a sandwich-like structure, based on the planar Kagome pattern, which has properties that make it attractive for application as a morphing material. In order to understand the passive and active properties of the KDLG with rigid joints, an analysis is made of the determinacy of the pin-jointed version. The number of internal mechanisms and states of self-stress of the finite pin-jointed structure are calculated as a function of the size of the structure. A statically and kinematically determinate version is obtained by relocating the internal nodes and by prescribing a set of patch bars around the periphery. The actuation performance of the rigid-jointed version is then explored theoretically by replacing a single bar in the structure by an actuator. The resistance to actuation is determined in terms of the stiffness and the allowable actuation strain as dictated by yield and buckling. The paper concludes with the optimal design of a double-layer grid to maximise actuation performance.     

Effective-Well Method for Matching and Predicting Oil Production

Decline curve analysis (DCA) is a traditional method for production prediction, which is still being used because it works in many cases and it is relatively simple to apply. However, DCA methods can neither match nor predict the fluctuating oil production during the early period when applied to entire reservoirs. The change in oil production may be because of variation in production conditions or the number of injection/production wells. In this study, we focused on the latter problem, change in the number of production wells. Obviously, there would be a significant oil production boost during a specific time period if more oil wells are drilled. The traditional DCA approach cannot match the increase in oil production due to the increase in the number of oil production wells. We have developed a method to match the oil production of entire reservoirs for the life span by considering the change in the number of production wells. The main idea of this approach came from the concept of effective wells. We applied this approach in several sandstone oil reservoirs with different permeabilities. The proposed effective-well model could match the oil production data in different reservoirs, even during the early period of production when the oil production rate change with time because of the variation in the number of producers. Comparison with the existing models (exponential, hyperbolic model, and harmonic models) was made and the results showed the proposed approach had the best fit to the production history in the cases studied.     

Effect of Well Orientation (Vertical vs. Horizontal) and Well Length on the Injection of CO<Subscript>2</Subscript> in Deep Saline Aquifers

Simulations of CO₂ injection into confined saline aquifers were conducted for both vertical and horizontal injection wells. The metrics used in quantifying the performances of different injection scenarios included changes in pressure near the injection well, mass of CO₂ dissolved into brine (solubility trapping), and storage efficiency, all evaluated with an assumed injection period of 50 years. Metrics were quantified as functions of well length, well orientation, CO₂ injection rate, and formation anisotropy (ratio of vertical to horizontal conductivity). When equal well lengths are compared, there is not a significant difference between the predicted performances of horizontal and vertical wells. However, the length of a horizontal well may exceed the length of a vertical well because the length of the horizontal well is not constrained to the vertical thickness of the geologic formation. Simulations show that, as the length of the horizontal well is allowed to increase, the geologic formation can receive a significantly higher CO₂ injection rate without exceeding a maximum allowable pressure. This result is observed in both isotropic and anisotropic formations, and suggests that horizontal wells may be advantageous under pressure-limited conditions. However, the use of horizontal wells does not significantly improve the storage efficiency, and under strongly anisotropic conditions, a vertical well provides higher storage efficiency than a horizontal well. We conclude that horizontal wells may be preferable if the goal is to sequester a large amount of CO₂ in a short period of time, but do not offer a significant advantage in terms of long-term capacity of a potential repository.     

Exact solutions of problems of steady fluid flow in reservoirs with hydrofractures

The plane steady flow of a homogeneous fluid between several wells in a reservoir zone containing hydrofractures is considered. Under the assumption of infinite fracture conductivity, analytic solutions of the Laplace equation for the pressure distribution in the elements of several typical symmetrical well arrangements are obtained, thus enabling them to be compared. Since the conductivity of the fracture may have a significant effect on the flow, the influence of its hydraulic properties on the productivity of a producing well is numerically investigated. Moscow. Translated from Izvestiya Rossiiskoi Akademii Nauk, Mekhanika Zhidkosti i Gaza, No. 2, pp. 74–82, March–April, 1994.     

Methods of calculating thermal conductivity of binary mixtures involving polyatomic gases

Summary The experimental binary thermal conductivity data of nineteen different gas pairs have been discussed and the competence of the rigorous, approximate and empirical procedures to represent them is investigated and discussed. In addition a new semi-theoretical method is suggested and tested. The suggested semi-theoretical procedure works very satisfactorily and is of good accuracy. It also compares favourably with the other methods. We also suggest a procedure for estimating thermal conductivity values at high temperature. This is an interesting and useful suggestion in view of the great practical need and their meagre availability.     

Scaling Mode Shapes in Output-Only Systems by a Consecutive Mass Change Method

In operational modal analysis (OMA) mode shapes can be obtained only with arbitrary normalization. There are many applications where mass normalized mode shapes are required, such as response prediction and stress analysis. A method to scale the mode shapes in OMA is to modify the dynamic behaviour of the structure by adding masses and then to use the modal parameters of both the original and modified structure. Several mass change methods have been proposed in recent years for estimating the scaling factors, where a distributed array of added masses are needed to obtain good results. In this work a new mass change approach based on performing several individual mass changes is presented. This approach requires only a small number of masses that are located at different points in each individual experiment. The results of the individual tests are then combined to estimate the scaling factors. The approach is developed and validated by measurements carried out on a 15-tonne prestressed concrete slab strip and a steel cantilever beam. The results show that a good accuracy can be obtained by this method when a proper mass change strategy is used.