Efficient identification of uncertain parameters in a large‐scale tidal model of the European continental shelf by proper orthogonal decomposition

The adjoint method can be used to identify uncertain parameters in large‐scale shallow water flow models. This requires the implementation of the adjoint model, which is a large programming effort. The work presented here is inverse modeling based on model reduction using proper orthogonal decomposition (POD). An ensemble of forward model simulations is used to determine the approximation of the covariance matrix of the model variability and the dominant eigenvectors of this matrix are used to define a model subspace. An approximate linear reduced model is obtained by projecting the original model onto this reduced subspace. Compared with the classical variational method, the adjoint of the tangent linear model is replaced by the adjoint of a linear reduced forward model. The minimization process is carried out in reduced subspace and hence reduces the computational costs. In this study, the POD‐based calibration approach has been implemented for the estimation of the depth values and the bottom friction coefficient in a large‐scale shallow sea model of the entire European continental shelf with approximately 10⁶ operational grid points. A number of calibration experiments is performed. The effectiveness of the algorithm is evaluated in terms of the accuracy of the final results as well as the computational costs required to produce these results. The results demonstrate that the POD calibration method with little computational effort and without the implementation of the adjoint code can be used to solve large‐scale inverse shallow water flow problems. Copyright © 2011 John Wiley & Sons, Ltd.     

Stochastic modeling and identification of an uncertain computational dynamical model with random fields properties and model uncertainties

This paper is devoted to the construction and to the identification of a probabilistic model of random fields in the presence of modeling errors, in high stochastic dimension and presented in the context of computational structural dynamics. Due to the high stochastic dimension of the random quantities which have to be identified using statistical inverse methods (challenging problem), a complete methodology is proposed and validated. The parametric–nonparametric (generalized) probabilistic approach of uncertainties is used to perform the prior stochastic models: (1) system-parameters uncertainties induced by the variabilities of the material properties are described by random fields for which their statistical reductions are still in high stochastic dimension and (2) model uncertainties induced by the modeling errors are taken into account with the nonparametric probabilistic approach in high stochastic dimension. For these two sources of uncertainties, the methodology consists in introducing prior stochastic models described with a small number of parameters which are simultaneously identified using the maximum likelihood method and experimental responses. The steps of the methodology are explained and illustrated through an application.     

Discharge estimation under uncertainty using variational methods with application to the full Saint‐Venant hydraulic network model

Estimating river discharge from in situ and/or remote sensing data is a key issue for evaluation of water balance at local and global scales and for water management. Variational data assimilation (DA) is a powerful approach used in operational weather and ocean forecasting, which can also be used in this context. A distinctive feature of the river discharge estimation problem is the likely presence of significant uncertainty in principal parameters of a hydraulic model, such as bathymetry and friction, which have to be included into the control vector alongside the discharge. However, the conventional variational DA method being used for solving such extended problems often fails. This happens because the control vector iterates (i.e., approximations arising in the course of minimization) result into hydraulic states not supported by the model. In this paper, we suggest a novel version of the variational DA method specially designed for solving estimation‐under‐uncertainty problems, which is based on the ideas of iterative regularization. The method is implemented with SIC², which is a full Saint‐Venant based 1D‐network model. The SIC² software is widely used by research, consultant and industrial communities for modeling river, irrigation canal, and drainage network behavior. The adjoint model required for variational DA is obtained by means of automatic differentiation. This is likely to be the first stable consistent adjoint of the 1D‐network model of a commercial status in existence. The DA problems considered in this paper are offtake/tributary estimation under uncertainty in the cross‐device parameters and inflow discharge estimation under uncertainty in the bathymetry defining parameters and the friction coefficient. Numerical tests have been designed to understand identifiability of discharge given uncertainty in bathymetry and friction. The developed methodology, and software seems useful in the context of the future Surface Water and Ocean Topography satellite mission. Copyright © 2016 John Wiley & Sons, Ltd.     

Adjoint‐based error estimation and mesh adaptation for hybridized discontinuous Galerkin methods

We present a robust and efficient target‐based mesh adaptation methodology, building on hybridized discontinuous Galerkin schemes for (nonlinear) convection–diffusion problems, including the compressible Euler and Navier–Stokes equations. The hybridization of finite element discretizations has the main advantage that the resulting set of algebraic equations has globally coupled degrees of freedom (DOFs) only on the skeleton of the computational mesh. Consequently, solving for these DOFs involves the solution of a potentially much smaller system. This not only reduces storage requirements but also allows for a faster solution with iterative solvers. The mesh adaptation is driven by an error estimate obtained via a discrete adjoint approach. Furthermore, the computed target functional can be corrected with this error estimate to obtain an even more accurate value. The aim of this paper is twofold: Firstly, to show the superiority of adjoint‐based mesh adaptation over uniform and residual‐based mesh refinement and secondly, to investigate the efficiency of the global error estimate. Copyright © 2014 John Wiley & Sons, Ltd.     

Convergence acceleration of continuous adjoint solvers using a reduced‐order model

A novel acceleration technique using a reduced‐order model is presented to speed up convergence of continuous adjoint solvers. The acceleration is achieved by projecting to an improved solution within an iterative process solely using early solution results. This is achieved by forming basis vectors from early iteration adjoint solutions to perform model order reduction of the adjoint equations. The reduced‐order model of the adjoint equations is then substituted into the full‐order discretized governing equations to determine weighting coefficients for each basis vector. With these coefficients, a linear combination of the basis vectors is used to project to an improved solution. The method is applied to 3 inviscid quasi‐1D nozzle flow cases including fully subsonic flow, subsonic inlet to supersonic outlet flow, and transonic flow with a shock. Significant cost reductions are achieved for a single application as well as repeated applications of the convergence acceleration technique.     

Sequential feature‐based mesh movement and adjoint error‐based mesh refinement

Nowadays, aerodynamic computational modeling is carried out on a daily basis in an industrial setting. This is done with the aim of predicting the performance and flow characteristics of new components. However, limited resources in terms of time and hardware force the engineer to employ relatively coarse computational grids, thus achieving results with variable degree of inaccuracy. In this article, a novel combination of feature and adjoint‐based mesh adaptation methods is investigated and applied to typical three‐dimensional turbomachinery cases, such as compressor and fan blades. The proposed process starts by employing feature‐based mesh movement to improve the global flow solution and then adjoint refinement to tune the mesh for each quantity of interest. Comparison of this process with one utilizing only the adjoint refinement procedure shows significant benefits in terms of accuracy of the performance quantity.     

Two‐dimensional prediction of time dependent,turbulent flow around a square cylinder confined in a channel

This paper presents two‐dimensional and unsteady RANS computations of time dependent, periodic, turbulent flow around a square block. Two turbulence models are used: the Launder–Sharma low‐Reynolds number k–ε model and a non‐linear extension sensitive to the anisotropy of turbulence. The Reynolds number based on the free stream velocity and obstacle side is Re=2.2×10⁴. The present numerical results have been obtained using a finite volume code that solves the governing equations in a vertical plane, located at the lateral mid‐point of the channel. The pressure field is obtained with the SIMPLE algorithm. A bounded version of the third‐order QUICK scheme is used for the convective terms. Comparisons of the numerical results with the experimental data indicate that a preliminary steady solution of the governing equations using the linear k–ε does not lead to correct flow field predictions in the wake region downstream of the square cylinder. Consequently, the time derivatives of dependent variables are included in the transport equations and are discretized using the second‐order Crank–Nicolson scheme. The unsteady computations using the linear and non‐linear k–ε models significantly improve the velocity field predictions. However, the linear k–ε shows a number of predictive deficiencies, even in unsteady flow computations, especially in the prediction of the turbulence field. The introduction of a non‐linear k–ε model brings the two‐dimensional unsteady predictions of the time‐averaged velocity and turbulence fields and also the predicted values of the global parameters such as the Strouhal number and the drag coefficient to close agreement with the data. Copyright © 2009 John Wiley & Sons, Ltd.     

A second‐order time accurate finite element method for quasi‐incompressible viscous flows

A finite element method for quasi‐incompressible viscous flows is presented. An equation for pressure is derived from a second‐order time accurate Taylor–Galerkin procedure that combines the mass and the momentum conservation laws. At each time step, once the pressure has been determined, the velocity field is computed solving discretized equations obtained from another second‐order time accurate scheme and a least‐squares minimization of spatial momentum residuals. The terms that stabilize the finite element method (controlling wiggles and circumventing the Babuska–Brezzi condition) arise naturally from the process, rather than being introduced a priori in the variational formulation. A comparison between the present second‐order accurate method and our previous first‐order accurate formulation is shown. The method is also demonstrated in the computation of the leaky‐lid driven cavity flow and in the simulation of a crossflow past a circular cylinder. In both cases, good agreement with previously published experimental and computational results has been obtained. Copyright © 2010 John Wiley & Sons, Ltd.     

Time‐dependent turbulent cavitating flow computations with interfacial transport and filter‐based models

Turbulent cavitating flow computations need to address both cavitation and turbulence modelling issues. A recently developed interfacial dynamics‐based cavitation model (IDCM) incorporates the interfacial transport into the computational modelling of cavitation dynamics. For time‐dependent flows, it is known that the engineering turbulence closure such as the original k–ε model often over‐predicts the eddy viscosity values reducing the unsteadiness. A recently proposed filter‐based modification has shown that it can effectively modulate the eddy viscosity, rendering better simulation capabilities for time‐dependent flow computations in term of the unsteady characteristics. In the present study, the IDCM along with the filter‐based k–ε turbulence model is adopted to simulate 2‐D cavitating flows over the Clark‐Y airfoil. The chord Reynolds number is Re=7.0 × 10⁵. Two angles‐of‐attack of 5 and 8° associated with several cavitation numbers covering different flow regimes are conducted. The simulation results are assessed with the experimental data including lift, drag and velocity profiles. The interplay between cavitation and turbulence models reveals substantial differences in time‐dependent flow results even though the time‐averaged characteristics are similar. Copyright © 2005 John Wiley & Sons, Ltd.     

On the incremental singular value decomposition method to support unsteady adjoint-based optimization

This paper proposes and evaluates an approximation model based on an incremental Singular Value Decomposition (iSVD) algorithm, for unsteady flow field reconstructions, needed for integrating the unsteady adjoint equations backward in time, within a gradient-based optimization loop. Due to the iSVD algorithm, the computational cost of solving the unsteady adjoint equations is reduced considerably, without practically affecting the accuracy of the computed gradient. Approximations to the unsteady flow fields are constructed while solving the time-varying flow equations (moving forward in time) and used to reconstruct these fields during the backward-in-time integration of the continuous adjoint equations. Optimization results obtained using the proposed method are compared to those computed using the binomial checkpointing technique, which acts as the reference method. Test cases for both flow control and shape optimization problems are presented.     

Numerical modeling of cohesive sediments dynamics in estuaries: Part I—Description of the model and simulations in the Po River Estuary

The present work contributes to the numerical modeling of complex turbulent multiphasic fluid flows occurring in estuarine channels. This research finds its motivation in the increasing need for efficient management of estuaries by taking into account the complex turbulent stratified flows encountered in estuaries and costal zones. A time‐dependent, 3D finite element model of suspended sediment transport taking into account the effects of cohesiveness between sediments is presented. The model estuary is the forced time‐dependent winds, time elevation at open boundaries and river discharge. To cope with the stiffness problems a decoupling method is employed to solve the shallow‐water equations of mass conservation, momentum and suspended sediment transport with the conventional hydrostatic pressure. The decoupling method partitions a time step into three subcycles according to the physical phenomena. In the first sub‐cycle the pure hydrodynamics including the k–ε turbulence model is solved, followed by the advection–diffusion equations for pollutants (salinity, temperature, suspended sediment concentration, (SSC)), and finally the bed evolution is solved. The model uses a mass‐preserving method based on the so‐called Raviart–Thomas finite element on the unstructured mesh in the horizontal plane, while the multi‐layers system is adopted in vertical with the conventional conforming finite element method, with the advantage that the lowermost and uppermost layers of variable height allow a faithful representation of the time‐varying bed and free surface, respectively. The model has been applied to investigate the SSC and seabed evolution in Po River Estuary (PRE) in Italy. The computed results mimic the field data well. Copyright © 2007 John Wiley & Sons, Ltd.     

Depth‐integrated,non‐hydrostatic model with grid nesting for tsunami generation,propagation, and run‐up

Tsunamis generated by earthquakes involve physical processes of different temporal and spatial scales that extend across the ocean to the shore. This paper presents a shock‐capturing dispersive wave model in the spherical coordinate system for basin‐wide evolution and coastal run‐up of tsunamis and discusses the implementation of a two‐way grid‐nesting scheme to describe the wave dynamics at resolution compatible to the physical processes. The depth‐integrated model describes dispersive waves through the non‐hydrostatic pressure and vertical velocity, which also account for tsunami generation from dynamic seafloor deformation. The semi‐implicit, finite difference model captures flow discontinuities associated with bores or hydraulic jumps through the momentum‐conserved advection scheme with an upwind flux approximation. The two‐way grid‐nesting scheme utilizes the Dirichlet condition of the non‐hydrostatic pressure and both the horizontal velocity and surface elevation at the inter‐grid boundary to ensure propagation of dispersive waves and discontinuities across computational grids of different resolution. The inter‐grid boundary can adapt to bathymetric features to model nearshore wave transformation processes at optimal resolution and computational efficiency. A coordinate transformation enables application of the model to small geographic regions or laboratory experiments with a Cartesian grid. A depth‐dependent Gaussian function smoothes localized bottom features in relation to the water depth while retaining the bathymetry important for modeling of tsunami transformation and run‐up. Numerical experiments of solitary wave propagation and N‐wave run‐up verify the implementation of the grid‐nesting scheme. The 2009 Samoa Tsunami provides a case study to confirm the validity and effectiveness of the modeling approach for tsunami research and impact assessment. Copyright © 2010 John Wiley & Sons, Ltd.     

Adjoint‐based airfoil optimization with discretization error control

In this article, we develop a new airfoil shape optimization algorithm based on higher‐order adaptive DG methods with control of the discretization error. Each flow solution in the optimization loop is computed on a sequence of goal‐oriented h‐refined or hp‐refined meshes until the error estimation of the discretization error in a flow‐related target quantity (including the drag and lift coefficients) is below a prescribed tolerance. Discrete adjoint solutions are computed and employed for the multi‐target error estimation and adaptive mesh refinement. Furthermore, discrete adjoint solutions are employed for evaluating the gradients of the objective function used in the CGs optimization algorithm. Furthermore, an extension of the adjoint‐based gradient evaluation to the case of target lift flow computations is employed. The proposed algorithm is demonstrated on an inviscid transonic flow around the RAE2822, where the shape is optimized to minimize the drag at a given constant lift and airfoil thickness. The effect of the accuracy of the underlying flow solutions on the quality of the optimized airfoil shapes is investigated. Copyright © 2014 John Wiley & Sons, Ltd.     

On the uncertainty of CFD in sail aerodynamics

A verification and validation procedure for yacht sail aerodynamics is presented. Guidelines and an example of application are provided. The grid uncertainty for the aerodynamic lift, drag and pressure distributions for the sails is computed. The pressures are validated against experimental measurements, showing that the validation procedure may allow the identification of modelling errors. Lift, drag and L₂ norm of the pressures were computed with uncertainties of the order of 1%. Convergence uncertainty and round‐off uncertainty are several orders of magnitude smaller than the grid uncertainty. The uncertainty due to the dimension of the computational domain is computed for a flat plate at incidence and is found to be significant compared with the other uncertainties. Finally, it is shown how the probability that the ranking between different geometries is correct can be estimated knowing the uncertainty in the computation of the value used to rank. Copyright © 2013 John Wiley & Sons, Ltd.     

A three‐dimensional numerical internal tidal model involving adjoint method

A three‐dimensional internal tidal model involving the adjoint method is constructed based on the nonlinear, time‐dependent, free‐surface hydrodynamic equations in spherical coordinates horizontally, and isopycnic coordinates vertically, subject to the hydrostatic approximations. This model consists of two submodels: the forward model is used for the simulation of internal tides, while the adjoint model is used for optimization of modal parameters. Mode splitting technique is employed in both forward and adjoint models. In this model, the adjoint method is employed to estimate model parameters by assimilating the interior observations. As a preliminary feasibility study, a set of ideal experiments with the model‐generated pseudo‐observations of surface currents are performed to invert the open boundary conditions (OBCs). In the ideal experiments, 14 kinds of bottom topographies and six kinds of predetermined distributions of OBCs are considered to examine their influence on experiment results. The inversion obtained satisfying results and all the predetermined distributions were successfully inverted. Analysis of results suggests the following: in the case where the spatial variation of the OBC distribution is great or the open boundary is close to a rough topography, the results will be comparatively poor, but still satisfactory; both the tidal elevations and currents can be simulated very accurately with the surface currents at several observation points; the assimilation precision could be reliable and able to reflect both of the inversion and simulation results in the whole field. The performance and results of ideal experiments give a preliminary indication that the construction of this model is successful. Copyright © 2011 John Wiley & Sons, Ltd.     

有界参数结构特征值的上下界定理

与方法近似性的结构特征值包含定理不同,给出参数近似性的结构的特征值上下界定理．在结构刚度矩阵和质量矩阵可以利用结构参数进行非员分解的条件下,通过区间分析,将特征值的上下界分解成两个广义特征值问题进行求解．结果可以看成是胡海昌教授的特征值质量包含定理和刚度包含定理在结构参数近似性特征值问题中的一种推广和应用．     

A unified approach to simulation and upscaling of single‐phase flow through vuggy carbonates

Numerical modeling of flow through vuggy porous media, mainly vuggy carbonates, is a challenging endeavor. Firstly, because the presence of vugs can significantly alter the effective porosity and permeability of the medium. Secondly, because of the co‐existence of porous and free flow regions within the medium and the uncertainties in defining the exact boundary between them. Traditionally, such heterogeneous systems are modeled by the coupled Darcy–Stokes equations. However, numerical modeling of flow through vuggy porous media using coupled Darcy–Stokes equations poses several numerical challenges particularly with respect to specification of correct interface condition between the porous and free‐flow regions. Hence, an alternative method, a more unified approach for modeling flows through vuggy porous media, the Stokes–Brinkman model, is proposed here. It is a single equation model with variable coefficients, which can be used for both porous and free‐flow regions. This also makes the requirement for interface condition redundant. Thus, there is an obvious benefit of using the Stokes–Brinkman equation, which can be reduced to Stokes or Darcy equation by the appropriate choice of parameters. At the same time, the Stokes–Brinkman equation provides a smooth transition between porous and free‐flow region, thereby taking care of the associated uncertainties. A numerical treatment for upscaling Stokes–Brinkman model is presented as an approach to use Stokes–Brinkman model for multi‐phase flow. Numerical upscaling methodology is validated by analyzing the error norm for numerical pressure convergence. Stokes–Brinkman single equation model is tested on a series of realistic data sets, and the results are compared with traditional coupled Darcy–Stokes model. Copyright © 2011 John Wiley & Sons, Ltd.     

Reducing memory requirements of unsteady adjoint by synergistically using check-pointing and compression

The unsteady adjoint method used in gradient-based optimization in 2D and, particularly, 3D industrial problems modeled by unsteady PDEs may have significant storage requirements and/or computational cost. The reason for this is that the backward in time integration of the adjoint equations requires the previously computed instantaneous flow fields to be available at each time-step. This article proposes remedies to this problem, by extending/upgrading relevant techniques proposed by the group of authors as well as other researchers. Their applicability is wide, even if these remedies are herein demonstrated in shape optimization problems in unsteady fluid mechanics. Check-pointing is in widespread use as it reduces the memory footprint and CPU cost of the optimization with a controllable computational overhead. Alternatively, flow field time-series can be stored in a lossless or lossly compressed form. The novelty of this article is the development of a Compressed Coarse-grained Check-Pointing strategy for second-order accurate schemes in time, by optimally combining check-pointing and lossy compression. The latter includes (a) the incremental Proper Generalized Decomposition (iPGD) algorithm and (b) a hybridization of the iPGD with the ZFP and Zlib algorithms. This is implemented within OpenFOAM, which is used to solve the flow and adjoint equations and conduct the optimization, and assessed in 2D/3D aerodynamic shape optimization problems on unstructured grids. Effectiveness in data reduction, computational cost, and reconstruction accuracy are compared, vis-à-vis also to the “standard” binomial check-pointing technique after adjusting it to second-order accurate schemes in time.