Journal of Engineering Mechanics

A change in temperature induces thermo-osmosis (T-O) and thermofiltration (T-F) effects in soils, which can influence consolidation under certain conditions. Whether such conditions are met is not always known a priori. Therefore, this paper first develops a coupled model to assess these effects on the thermohydromechanical coupled behavior of saturated soils. The proposed model, which incorporates the extended Darcy’s law and Fourier’s law, is implemented in the finite-element method (FEM) solver OpenGeoSys version 6.3.3. To determine when T-O/T-F effects become relevant in an analysis, two characteristic numbers are then derived from the extended governing equations, which can be directly expressed by readily available soil and water parameters. Both theoretical and numerical analyses were conducted, and the results indicate that the T-F effect has little influence on heat transfer in typical soils, while the T-O effect appears to be significant for both sandy and clayey soils, provided the T-O coefficient exceeds $4.3\times {10}^{-12}{\mathrm{m}}^2/(\mathrm{s}·\mathrm{K})$. The T-O effect could even lead to the development of negative excess pore water pressure (PWP) and, consequently, soil compression, with relevance for the stability of geotechnical facilities subjected to temperature gradients. It is thus inferred that the T-O coefficient should be measured in conjunction with related necessary experimental work, in addition to soil permeability, to determine whether the T-O effect should be treated with caution.

Water or steam injection in oil fields is a common method for enhanced oil recovery in petroleum engineering. The thermoviscous fingering instability is one of the main problems of a complex nature that decreases the efficiency of oil extraction. Actually, oil wells are a porous medium with a level of anisotropy for permeability and diffusion. In this paper, the thermal viscous fingering instability in anisotropic media is investigated using both linear stability analysis and computational fluid dynamics (CFD) simulation. For stability analysis, the growth rate of disturbances is determined by solving quasi-steady-state equations via a shooting method. The CFD simulation is performed by solving the governing equations of heat and mass transfer using a spectral method. It is shown that the longitudinal direction permeability and the transverse direction dispersion have important effects on the instability. The values of the thermal-lag coefficient and the Lewis number have opposite effects on the different types of displacements considered. When porous media are swept using cold fluid, increasing the Lewis number intensifies the level of flow instability.

Fluctuating surface pressures on a bluff body exposed to a boundary layer flow generally are characterized as a spatiotemporally varying random field. In this paper, a dynamic mode decomposition (DMD) was applied to extract dominant features embedded in these random pressure fields. Utilizing an unsupervised machine learning algorithm, spatial modes and their temporal variations were grouped into different clusters at scales, e.g., macro, meso, and micro. A proper orthogonal decomposition (POD) of the experimental data was carried out to observe commonalities and distinctive perspectives each decomposition offers. A comprehensive examination of the DMD/POD for their convergence criteria, data sufficiency, and modal components analysis was conducted. The physical interpretation of the spatiotemporal pressure field based on these decomposition schemes was discussed. At different scales, the DMD modes can capture the evolution of aerodynamic features, e.g., convection of vortices (or vortex tubes) and other structures. The distribution of energy among these three broad scales also reflects an energy cascade in pressure fluctuations akin to turbulence.

The nonlinear problem of an inflated, partially-floating membrane cylinder with a bottom load is studied by perturbation and numerical integration. The solutions can be classified into stages as the load is increased. Stage I does not collapse anywhere. Stage II is still floating, partially collapses at the bottom, and shows semisimilarity. Stage III is completely immersed with partial collapse and shows full similarity. There exists a maximum load for Stage I. Stages II and III are unstable to constant load but can be realized by an anchored tether.

Tunneling stoppage often occurs during tunnel construction, especially in weak and rheological ground. The discontinuous excavation caused by the stoppage usually has a certain impact on the tunnel convergence and supporting pressure. This study develops a closed-form analytical solution for a deep-buried circular tunnel excavated in viscoelastic rock, in which the progress of longitudinal discontinuous excavation (the stoppage that occurs during tunneling) and the sequential installation of double elastic liners are considered. In the derivation, the change in the advance rate of the tunnel face before and after stopping is also considered. Corresponding to the whole construction process, the solution is solved step by step in five stages, and a generalized derivation procedure for any viscoelastic model is provided depending on the integral equation theory. Particularly for the Burgers model, explicit expressions for the time-dependent rock displacement and stress as well as the pressure on the two liners are presented, which can be reduced to the previous viscoelastic solutions with a constant advance rate or without the face effect. For validation purpose, comparisons are made between the proposed solution and previous existing solutions as well as the numerical results obtained by the finite-difference simulations. Additionally, to further verify the applicability of the presented solution in tunnel applications, the time-dependent tunnel convergence and liner stress of the four typical cross sections in the Saint Martin La Porte access adit are predicted, which are in good agreement with the corresponding field monitoring data. Compared with previous solutions, the proposed solution is more comprehensive and suitable for more complex tunnels with various excavation methods, rheological rocks, and supporting forms.

This paper presents a full-waveform inversion method for reconstructing the temporal and spatial distribution of unknown, incoherent dynamic traction in a heterogeneous, bounded solid domain from sparse, surficial responses. This work considers SH wave motions in a two-dimensional (2D) domain. The partial differential equation (PDE)-constrained optimization framework is employed to search a set of control parameters, by which a misfit between measured responses at sensors on the top surface induced by targeted traction and their computed counterparts induced by estimated traction is minimized. To mitigate the solution multiplicity of the presented inverse problem, we employ the Tikhonov (TN) regularization on the estimated traction function. We present the mathematical modeling and numerical implementation of both optimize-then-discretize (OTD) and discretize-then-optimize (DTO) approaches. The finite-element method (FEM) is employed to obtain the numerical solutions of state and adjoint problems. Newton’s method is utilized for estimating an optimal step length in combination with the conjugate-gradient scheme, calculating a desired search direction, throughout a minimization process. Numerical results present that the complexity of a material profile in a domain increases the error between reconstructed traction and its target. Second, the OTD and DTO approaches lead to the same inversion result. Third, when the sampling rate of the measurement is equal to the time step for discretizing estimated traction, the ratio of the size of measurement data to the number of the control parameters can be as small as $1:12$ in the presented work. Fourth, it is acceptable to tackle the presented inverse modeling of dynamic traction without the TN regularization. Fifth, the inversion performance is more compromised when the noise of a larger level is added to the measurement data, and using the TN regularization does not improve the inversion performance when noise is added to the measurement. Sixth, our minimizer suffers from solution multiplicity less when it identifies dynamic traction of lower frequency content than that of higher frequency content. The wave responses in a computational domain, induced by targeted traction and its reconstructed one, are in excellent agreement with each other. Thus, if the presented dynamic-input inversion algorithm is extended in realistic 3D settings, it could reconstruct seismic-input motions in a truncated domain and, then, replay the wave responses in a computational domain.

A new coupled damage-plastic constitutive model is presented for modeling the nonlinear behavior of concrete in the small strain space. The active damage tensor concept and the closure effect are presented by considering both the state of the strain and the boundary condition of the crack. A new Helmholtz free energy function (HFEF) is introduced with the existence of a dissipation plastic potential in the stain space. The staggering process between the plastic and damage states has been developed through the HFEF formulation and simultaneously satisfies the consistency conditions. The plastic calculations are formulated in the effective stress regime by developing a 3D plasticity model. The damage surface and damage evolution are formulated based on the accumulated plastic history. The numerical results from the proposed model have been compared with both the experimental data from the literature and with another numerical damage model. Finally, the model exhibits its ability to be applied to analyze various concrete structures.

A nonlinear dynamic framework for the analysis of beam–foundation interaction is developed. The extended Hamilton’s principle is used to analyze the response of Euler–Bernoulli beams vibrating on nonlinear continuums (e.g., soils) and subjected to vibrating and moving loads. The continuum beneath the beam is characterized by a nonlinear elastic constitutive relationship that connects the secant shear modulus to the induced strain. The novel feature of the analysis is that the nonlinear compression and shear parameters $k_s$ and $t_s$ of the continuum (i.e., foundation) do not have to be assumed a priori, as is required in conventional beams on foundation analysis, and are obtained as part of the solution. In fact, it is shown that these parameters are not constant but change with time and depend on the beam–foundation interaction. Another novel feature of the analysis is that the mass of the foundation participating in the vibration is obtained as part of the solution and does not have to be assumed a priori. Thus, the analysis rigorously takes into account the nonlinear beam–foundation (soil–structure) interaction within a dynamic time-integration framework. The developed framework is as accurate as and about 50% faster than conventional nonlinear dynamic finite element analysis. Inputs to the analysis can be given in the form of a text file without any requirement of numerical mesh generation, making the approach rather user friendly. The characteristics of the developed nonlinear dynamic foundation model are illustrated through examples of moving and vibrating loads.

This study proposes iterative methods for updating damping and stiffness matrices of a finite-element model based on measured complex eigenvalues, i.e., poles, which could have been obtained accurately by using various experimental modal analysis (EMA) or operational modal analysis (OMA) techniques. In each iteration of the proposed methods, updating the stiffness and damping parameters requires solving a set of simultaneous linear equations, which can be generated by following a traditional sensitivity-based (SB) method or a less-known cross-model cross-mode (CMCM) approach. Whereas the SB method uses the difference between the target and baseline models to generate equations, the CMCM-based approach utilizes the product between these two models. Analytically, the similarities and differences between them are fully explained. Through the numerical studies, it reveals that the CMCM-based approach might significantly outperform the SB method.

This paper presents the development, testing, and characterization of an innovative low-cost fail-safe sustainable energy-dissipation device in which the material surrounding the moving piston and enclosed within the damper housing is pressurized sand. The proposed sand damper does not suffer from the challenge of viscous heating and failure of its end seals, and it can be implemented in harsh environments with either high or low temperatures. Its symmetric force output is velocity-independent, and it can be continuously monitored and adjusted at will with standard commercially available strain gauges installed along the post-tensioned rods that exert the pressure on the sand. Component testing at various levels of pressure, stroke amplitude, and cycling frequency show that the proposed pressurized sand damper exhibits stable hysteretic cyclic behavior with increasing pinching at larger strokes. The paper examines the fidelity of an eight-parameter Bouc-Wen hysteretic model capable to model pinching and concludes that the proposed hysteretic model is able to capture the pronounced pinching of the hysteretic behavior at larger stroke amplitudes. Four of the eight parameters of the proposed hysteretic model can be determined a priori from physical arguments; therefore, only the remaining four parameters need to be determined from nonlinear regression analysis.

The Coriolis effect in centrifuge modeling of flow–structure interaction has not been well understood, and thus this paper presents numerical simulations to address this issue. The results indicate that the Coriolis acceleration obviously regulated the flow–structure interaction pattern. The maximum total impact force was amplified by approximately 2 times, and the force was distributed along the entire barrier surface when changing the Coriolis acceleration direction from acting at the slope to acting away from the slope. Reducing the flow velocity by 30%–40% decreased the amplification ratio by approximately 14%–29%. The alteration of the microcontact condition and the energy consumption of the dry granular flow was the main influence mechanism of the Coriolis acceleration on the flow’s impact behavior. The influence of the Coriolis effect on the viscous flow impact was completely different from that exerted on frictional flows. This discrepancy resulted from the intrinsic flow mobility determined by the material characteristics. Some practical discussions about centrifuge modeling of flow–structure interaction are made as well as some suggestions for future work.

This study aims to investigate the effects of the dynamic increase factor for beam structures under conventional blast loading. Combined with differential equations of an equivalent single-degree-of-freedom (SDOF) system in elastic and plastic response stages, the dynamic increase factor expression of a beam structure under conventional blast loading was derived. Based on the degree to which the structure reached the plastic stage, the dynamic increase factors under different ratios of equivalent mass-load transformation factors in the elastic stage and elastoplastic stage were analyzed. Using simply supported beams, simply supported and fixed beams, and fixed-end beams as the types of analysis beams, nine calculation conditions of the dynamic increase factor were completed. The calculation results were compared with the formula used by the Chinese blast-resistant design code. The results showed that for a beam structure design with a large ductility ratio $\beta$ ($\beta >2$), a ratio of the equivalent mass-load transformation factors in the elastic and plastic stages larger than 1 would increase engineering costs. Compared with simply supported beams and simply fixed beams, the relative error caused by the different ratios for the equivalent mass-load transformation factors is smaller for fixed-end beams. The existing research ignores the equivalent mass-load transformation factor in the plastic stage and takes the ratio of the equivalent mass-load transformation factors in the elastic and plastic stages as 1. This will have a significant impact on the dynamic increase factor of simply supported beams, and the value of maximum error is approximately 18%. If the equivalent mass-load transformation factor in the plastic stage is ignored, then the error in calculating the dynamic increase factor will be small for beam structures with lower ductility ratios $\beta$ ($\beta <1.6$), and the value of maximum error is approximately 5%.