Adjoint-based estimation and control of spatial, temporal and stochastic approximation errors in unsteady flow simulations
Not Available
SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Not Available
Schemes based on discrete adjoint algorithms present several exciting opportunities for significantly advancing the current state of the art in computational fluid dynamics. Such methods provide an extremely efficient means for obtaining discretely consistent sensitivity information for hundreds of design variables, opening the door to rigorous, automated design optimization of complex aerospace configuration using the Navier-Stokes equation. Moreover, the discrete adjoint formulation provides a mathematically rigorous foundation for mesh adaptation and systematic reduction of spatial discretization error. Error estimates are also an inherent by-product of an adjoint-based approach, valuable information that is virtually non-existent in today's large-scale CFD simulations. An overview of the adjoint-based algorithm work at NASA Langley Research Center is presented, with examples demonstrating the potential impact on complex computational problems related to design optimization as well as mesh adaptation.
This project addressed the challenge of predictive computational analysis of strongly coupled, highly nonlinear multiphysics systems characterized by multiple physical phenomena that span a large range of length- and time-scales. Specifically, the project was focused on computational estimation of numerical error and sensitivity analysis of computational solutions with respect to variations in parameters and data. In addition, the project investigated the use of accurate computational estimates to guide efficient adaptive discretization. The project developed, analyzed and evaluated new variational adjoint-based techniques for integration, model, and data error estimation/control and sensitivity analysis, in evolutionary multiphysics multiscale simulations.
Objectives: Reliable evaluation of mass flow rates through permeable boundaries - Estimate and control discretization error- Consider both computational domain outflow and inflow- Applicable to simulating propulsion-system effects, as well as secondary flow paths - Explore feasibility of handling more general outputs at domain boundaries. Design optimization subject to mass-flow-rate constraints - Improve aerodynamic performance and reduce noise due to sonic boom - Control discretization error in design space to improve confidence in final designs.
Engineering problems commonly require functional outputs of computational fluid dynamics (CFD) simulations with specified accuracy. These simulations are performed with limited computational resources. Computable error estimates offer the possibility of quantifying accuracy on a given mesh and predicting a fine grid functional on a coarser mesh. Such an estimate can be computed by solving the flow equations and the associated adjoint problem for the functional of interest. An adjoint-based error correction procedure is demonstrated for transonic inviscid and subsonic laminar and turbulent flow. A mesh adaptation procedure is formulated to target uncertainty in the corrected functional and terminate when error remaining in the calculation is less than a user-specified error tolerance. This adaptation scheme is shown to yield anisotropic meshes with corrected functionals that are more accurate for a given number of grid points then isotropic adapted and uniformly refined grids.
A technique is developed to answer the important question: 'Given limited system response measurements and ever-present physical limits on the level of excitation, what excitation should be provided to a system to make damage most detectable?' Specifically, a method is presented for optimizing excitations that maximize the sensitivity of output measurements to perturbations in damage-related parameters estimated with an extended Kalman filter. This optimization is carried out in a computationally efficient manner using adjoint-based optimization and causes the innovations term in the extended Kalman filter to be larger in the presence of estimation errors, which leads to a better estimate of the damage-related parameters in question. The technique is demonstrated numerically on a nonlinear 2 DOF system, where a significant improvement in the damage-related parameter estimation is observed.
Engineering computational fluid dynamics (CFD) analysis and design applications focus on output functions (e.g., lift, drag). Errors in these output functions are generally unknown and conservatively accurate solutions may be computed. Computable error estimates can offer the possibility to minimize computational work for a prescribed error tolerance. Such an estimate can be computed by solving the flow equations and the linear adjoint problem for the functional of interest. The computational mesh can be modified to minimize the uncertainty of a computed error estimate. This robust mesh-adaptation procedure automatically terminates when the simulation is within a user specified error tolerance. This procedure for estimating and adapting to error in a functional is demonstrated for three-dimensional Euler problems. An adaptive mesh procedure that links to a Computer Aided Design (CAD) surface representation is demonstrated for wing, wing-body, and extruded high lift airfoil configurations. The error estimation and adaptation procedure yielded corrected functions that are as accurate as functions calculated on uniformly refined grids with ten times as many grid points.
Adjoint solutions of the governing flow equations are becoming increasingly important for the development of efficient analysis and optimization algorithms. A well-known use of the adjoint method is gradient-based shape optimization. Given an objective function that defines some measure of performance, such as the lift and drag functionals, its gradient is computed at a cost that is essentially independent of the number of design variables (geometric parameters that control the shape). More recently, emerging adjoint applications focus on the analysis problem, where the adjoint solution is used to drive mesh adaptation, as well as to provide estimates of functional error bounds and corrections. The attractive feature of this approach is that the mesh-adaptation procedure targets a specific functional, thereby localizing the mesh refinement and reducing computational cost. Our focus is on the development of adjoint-based optimization techniques for a Cartesian method with embedded boundaries.12 In contrast t o implementations on structured and unstructured grids, Cartesian methods decouple the surface discretization from the volume mesh. This feature makes Cartesian methods well suited for the automated analysis of complex geometry problems, and consequently a promising approach to aerodynamic optimization. Melvin et developed an adjoint formulation for the TRANAIR code, which is based on the full-potential equation with viscous corrections. More recently, Dadone and Grossman presented an adjoint formulation for the Euler equations. In both approaches, a boundary condition is introduced to approximate the effects of the evolving surface shape that results in accurate gradient computation. Central to automated shape optimization algorithms is the issue of geometry modeling and control. The need to optimize complex, "real-life" geometry provides a strong incentive for the use of parametric-CAD systems within the optimization procedure. In previous work, we presented an effective optimization framework that incorporates a direct-CAD interface. In this work, we enhance the capabilities of this framework with efficient gradient computations using the discrete adjoint method. We present details of the adjoint numerical implementation, which reuses the domain decomposition, multigrid, and time-marching schemes of the flow solver. Furthermore, we explain and demonstrate the use of CAD in conjunction with the Cartesian adjoint approach. The final paper will contain a number of complex geometry, industrially relevant examples with many design variables to demonstrate the effectiveness of the adjoint method on Cartesian meshes.
We use 2009–2011 space-borne methane observations from the Greenhouse Gases Observing SATellite (GOSAT) to estimate global and North American methane emissions with 4° × 5° and up to 50 km × 50 km spatial resolution, respectively. GEOS-Chem and GOSAT data are first evaluated with atmospheric methane observations from surface and tower networks (NOAA/ESRL, TCCON) and aircraft (NOAA/ESRL, HIPPO), using the GEOS-Chem chemical transport model as a platform to facilitate comparison of GOSAT with in situ data. This identifies a high-latitude bias between the GOSAT data and GEOS-Chem that we correct via quadratic regression. Our global adjoint-based inversion yields a total methane source of 539 Tg a −1 with some important regional corrections to the EDGARv4.2 inventory used as a prior. Results serve as dynamic boundary conditions for an analytical inversion of North American methane emissions using radial basis functions to achieve high resolution of large sources and provide error characterization. We infer a US anthropogenic methane source of 40.2–42.7 Tg a −1 , as compared to 24.9–27.0 Tg a −1 in the EDGAR and EPA bottom-up inventories, and 30.0–44.5 Tg a −1 in recent inverse studies. Our estimate is supported by independent surface and aircraft data and by previous inverse studies for California. We find that the emissions are highest in the southern–central US, the Central Valley of California, and Florida wetlands; large isolated point sources such as the US Four Corners also contribute. Using prior information on source locations, we attribute 29–44 % of US anthropogenic methane emissions to livestock, 22–31 % to oil/gas, 20 % to landfills/wastewater, and 11–15 % to coal. Wetlands contribute an additional 9.0–10.1 Tg a −1 .
Summary In this study, we demonstrate geomechanical modeling with fully automatic parameter calibration to estimate the full geomechanical stress fields of a prospective US carbon dioxide (CO2) storage site, based on sparse measurement data. The goal is to compute full stress tensor field estimates (principal stresses and orientations) that are maximally compatible with observations within the constraints of the model assumptions, thereby extending pointwise, incomplete partial stress measurement to a simulated full formation stress field, as well as a rough assessment of the associated error. We use the Perch site, located in Otsego County, Michigan, USA, as our case study. The input data consist of partial stress tensor information inferred from in-situ borehole tests, geophysical well logs, and processing of seismic data. A static earth model (SEM) of the site was developed, and geomechanical simulation functionality of the open-source MATLAB Reservoir Simulation Toolbox (MRST) was used to model the stress field. Adjoint-based nonlinear optimization was used to adjust boundary conditions and material properties to calibrate simulated results of observations. Results were interpreted through a Bayesian framework. The focus of this paper is to demonstrate how the fully automatic calibration procedure works and discuss the results obtained; it does not attempt a detailed analysis of the stress field in the context of the proposed CO2 storage initiatives. Our work is part of a larger effort to noninvasively determine in-situ stresses in deep formations considered for CO2 storage. Guided by previously published research on geomechanical model calibration, our work presents a novel calibration approach supporting a potentially large number of linear or nonlinear calibration parameters to produce results optimally agreeing with available measurements and thus extend partial pointwise estimates to full tensor fields compatible with the physics of the site.
We present a scalable, parallel mechanism for stochastic identification/control for problems constrained by partial differential equations with random input data. Several identification objectives will be discussed that either minimize the expectation of a tracking cost functional or minimize the difference of desired statistical quantities in the appropriate $L^p$ norm, and the distributed parameters/control can both deterministic or stochastic. Given an objective we prove the existence of an optimal solution, establish the validity of the Lagrange multiplier rule and obtain a stochastic optimality system of equations. The modeling process may describe the solution in terms of high dimensional spaces, particularly in the case when the input data (coefficients, forcing terms, boundary conditions, geometry, etc) are affected by a large amount of uncertainty. For higher accuracy, the computer simulation must increase the number of random variables (dimensions), and expend more effort approximating the quantity of interest in each individual dimension. Hence, we introduce a novel stochastic parameter identification algorithm that integrates an adjoint-based deterministic algorithm with the sparse grid stochastic collocation FEM approach. This allows for decoupled, moderately high dimensional, parameterized computations of the stochastic optimality system, where at each collocation point, deterministic analysis and techniques can be utilized. The advantage of our approach is that it allows for the optimal identification of statistical moments (mean value, variance, covariance, etc.) or even the whole probability distribution of the input random fields, given the probability distribution of some responses of the system (quantities of physical interest). Our rigorously derived error estimates, for the fully discrete problems, will be described and used to compare the efficiency of the method with several other techniques. Numerical examples illustrate the theoretical results and demonstrate the distinctions between the various stochastic identification objectives.
Observing system simulation experiments (OSSEs) are powerful tools for estimating the usefulness of various configurations of envisioned observing systems and data assimilation techniques. Their utility stems from their being conducted in an entirely simulated context, utilizing simulated observations having simulated errors and drawn from a simulation of the earth's environment. Observations are generated by applying physically based algorithms to the simulated state, such as performed during data assimilation or using other appropriate algorithms. Adding realistic instrument plus representativeness errors, including their biases and correlations, can be critical for obtaining realistic assessments of the impact of a proposed observing system or analysis technique. If estimates of the expected accuracy of proposed observations are realistic, then the OSSE can be also used to learn how best to utilize the new information, accelerating its transition to operations once the real data are available. As with any inferences from simulations, however, it is first imperative that some baseline OSSEs are performed and well validated against corresponding results obtained with a real observing system. This talk provides an overview of, and highlights critical issues related to, the development of an OSSE framework for the tropospheric weather prediction component of the NASA GEOS-5 global atmospheric data assimilation system. The framework includes all existing observations having significant impact on short-term forecast skill. Its validity has been carefully assessed using a range of metrics that can be evaluated in both the OSSE and real contexts, including adjoint-based estimates of observation impact. A preliminary application to the Aeolus Doppler wind lidar mission, scheduled for launch by the European Space Agency in 2014, has also been investigated.
An adjoint solver is implemented in the FUN3D stabilized finite-element flow solver. The adjoint solution is used to generate anisotropic, adapted meshes to control error in scalar out- put functionals, such as lift or drag coefficient. The inviscid and turbulent adjoints are verified with a finite-difference derivative approximation and can be used in design optimization in addition to mesh adaptation. The adjoint capability represents an extension of previous research using the FUN3D stabilized finite-element capability with metric-based mesh adaptation and interpolation-based error estimates to generate highly anisotropic adapted meshes for turbulent flows. In the present work, a metric-based approach is again utilized, where the adjoint and the primal solutions both contribute to the generation of a metric tensor field that is subsequently used to produce the required anisotropic mesh for each adaptation cycle. Adaptive results are then shown for an inviscid supersonic flow over a diamond airfoil, inviscid transonic flow over the ONERA M6 wing, and viscous laminar flow over the NACA 0012 airfoil, all using drag force as the output functional. Adjoint-based adaptation is compared with a multiscale solution-based approach that controls the L norm of Mach number interpolation error to demonstrate the effectiveness and effciency of the adjoint-based adaptive mesh technology.
A critical aspect of applying modern computational solution methods to complex multiphysics systems of relevance to nuclear reactor modeling, is the assessment of the predictive capability of specific proposed mathematical models. In this respect the understanding of numerical error, the sensitivity of the solution to parameters associated with input data, boundary condition uncertainty, and mathematical models is critical. Additionally, the ability to evaluate and or approximate the model efficiently, to allow development of a reasonable level of statistical diagnostics of the mathematical model and the physical system, is of central importance. In this study we report on initial efforts to apply integrated adjoint-based computational analysis and automatic differentiation tools to begin to address these issues. The study is carried out in the context of a Reynolds averaged Navier–Stokes approximation to turbulent fluid flow and heat transfer using a particular spatial discretization based on implicit fully-coupled stabilized FE methods. Initial results are presented that show the promise of these computational techniques in the context of nuclear reactor relevant prototype thermal-hydraulics problems.
During 2016–17, the Juno and Cassini spacecraft will both perform close eccentric orbits of Jupiter and Saturn, respectively, obtaining high-precision gravity measurements for these planets. These data will be used to estimate the depth of the observed surface flows on these planets. All models to date, relating the winds to the gravity field, have been in the forward direction, thus only allowing the calculation of the gravity field from given wind models. However, there is a need to do the inverse problem since the new observations will be of the gravity field. Here, an inverse dynamical model is developed to relate the expected measurable gravity field, to perturbations of the density and wind fields, and therefore to the observed cloud-level winds. In order to invert the gravity field into the 3D circulation, an adjoint model is constructed for the dynamical model, thus allowing backward integration. This tool is used for the examination of various scenarios, simulating cases in which the depth of the wind depends on latitude. We show that it is possible to use the gravity measurements to derive the depth of the winds, both on Jupiter and Saturn, also taking into account measurement errors. Calculating the solution uncertainties, we show that the wind depth can be determined more precisely in the low-to-mid-latitudes. In addition, the gravitational moments are found to be particularly sensitive to flows at the equatorial intermediate depths. Therefore, we expect that if deep winds exist on these planets they will have a measurable signature by Juno and Cassini.
The mesh adaptation functionality of FUN3D is utilized to obtain a mesh optimized to calculate sonic boom ground signature loudness. During this process, the coupling between the discrete-adjoints of the computational fluid dynamics tool FUN3D and the atmospheric propagation tool sBOOM is exploited to form the error estimate. This new mesh adaptation methodology will allow generation of suitable meshes adapted to reduce the estimated errors in the ground loudness, which is an optimization metric employed in supersonic aircraft design. This new output-based adaptation could allow new insights into meshing for sonic boom analysis and design, and complements existing output-based adaptation techniques such as adaptation to reduce estimated errors in off-body pressure functional. This effort could also have implications for other coupled multidisciplinary adjoint capabilities (e.g., aeroelasticity) as well as inclusion of propagation specific parameters such as prevailing winds or non-standard atmospheric conditions. Results are discussed in the context of existing methods and appropriate conclusions are drawn as to the efficacy and efficiency of the developed capability.
Adjoints of numerical weather prediction models may be employed for Forecast Sensitivity to Observation (FSO) in order to monitor the contribution of ingested observation data on short-term forecast skill. However, the calculation of short-term forecast error is difficult due to the lack of a truly independent dataset for verification. In an Observing System Simulation Experiment framework, the Nature Run is able to provide a true and complete verification dataset and allows accurate evaluation of short term forecast errors. In this work, an OSSE developed at the National Aeronautics and Space Administration Global Modeling and Assimilation Office is used to explore the impact of observational data on forecasts in the 6 to 48 hour range. An adjoint of the Global Earth Observing System model is employed to compare the observation impacts estimated using both self-analysis verification and the true Nature Run verification. Self-analysis verification is found to inflate the estimated forecast error growth during the early forecast period, resulting in overestimations of observation impacts, particularly in the 6-12 hour forecast range. By 48 hours, the self-analysis verification estimates of forecast error and observation impacts more closely match the true values. The fraction of beneficial observations is also overinflated at short forecast times when self-analysis verification is used. The progression of impacts of an individual observation or data type depends on the character of the growth of the initial condition error that each observation affects.
Simulation results are presented for all cases from the Third AIAA Sonic Boom Prediction Workshop. An inviscid, embedded-boundary Cartesian-mesh flow solver is used in conjunction with adjoint-based mesh adaptation to compute nearfield pressure signatures. Specialized techniques are applied to maximize accuracy and minimize cost on Cartesian meshes. The Richardson-based error estimate highlights regions of the signatures most sensitive to mesh refinement. Timing results and coarse, medium, and fine mesh sizes for nearfield cases demonstrate that the parallel decomposition approach is efficient in both computational time and wall-clock. Pressure signals are propagated to the ground using an augmented Burgers’ equation solver to predict boom carpets. Ground signatures and loudness metrics are presented for a standard atmosphere as well as more realistic atmospheric profiles, which affect overall noise levels and can significantly widen the boom carpet. Mesh convergence studies show that high sampling frequencies, around 500 kHz, are required for propagation, and the sampling frequency increases at large off-track angles with longer acoustic ray paths and propagation times. The numerical methods yield accurate results for predicting low sonic boom signatures while being among the least computationally expensive of the workshop.