Forward integration and Adjoint Optimization for Dynamic Coupled Scales
Explore the source record for details and available documents.
97 MATHEMATICS AND COMPUTING↗
SEARCH · Engineering Papers
Results for “adjoint optimization”
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.
Trajectory Optimization Using Adjoint Method and Chebyshev Polynomial Approximation for Minimizing Fuel Consumption During Climb
This paper describes two methods of trajectory optimization to obtain an optimal trajectory of minimum-fuel- to-climb for an aircraft. The first method is based on the adjoint method, and the second method is based on a direct trajectory optimization method using a Chebyshev polynomial approximation and cubic spine approximation. The approximate optimal trajectory will be compared with the adjoint-based optimal trajectory which is considered as the true optimal solution of the trajectory optimization problem. The adjoint-based optimization problem leads to a singular optimal control solution which results in a bang-singular-bang optimal control.
Chebyshev Polynomial↗
An adjoint-based optimization method for jointly inverting heterogeneous material properties and fault slip from earthquake surface deformation data
SUMMARY Analysis of tectonic and earthquake-cycle associated deformation of the crust can provide valuable insights into the underlying deformation processes including fault slip. How those processes are expressed at the surface depends on the lateral and depth variations of rock properties. The effect of such variations is often tested by forward models based on a priori geological or geophysical information. Here, we first develop a novel technique based on an open-source finite-element computational framework to invert geodetic constraints directly for heterogeneous media properties. We focus on the elastic, coseismic problem and seek to constrain variations in shear modulus and Poisson’s ratio, proxies for the effects of lithology and/or temperature and porous flow, respectively. The corresponding nonlinear inversion is implemented using adjoint-based optimization that efficiently reduces the cost function that includes the misfit between the calculated and observed displacements and a penalty term. We then extend our theoretical and numerical framework to simultaneously infer both heterogeneous Earth’s structure and fault slip from surface deformation. Based on a range of 2-D synthetic cases, we find that both model parameters can be satisfactorily estimated for the megathrust setting-inspired test problems considered. Within limits, this is the case even in the presence of noise and if the fault geometry is not perfectly known. Our method lays the foundation for a future reassessment of the information contained in increasingly data-rich settings, for example, geodetic GNSS constraints for large earthquakes such as the 2011 Tohoku-oki M9 event, or distributed deformation along plate boundaries as constrained from InSAR.
Geochemistry & Geophysics↗
Enhanced Data Efficiency Using Deep Neural Networks and Gaussian Processes for Aerodynamic Design Optimization
Adjoint-based optimization methods are attractive for aerodynamic shape design primarily due to their computational costs being independent of the dimensionality of the input space and their ability to generate high-fidelity gradients that can then be used in a gradient-based optimizer. This makes them very well suited for high-fidelity simulation based aerodynamic shape optimization of highly parametrized geometries such as aircraft wings. However, the development of adjoint-based solvers involve careful mathematical treatment and their implementation require detailed software development. Furthermore, they can become prohibitively expensive when multiple optimization problems are being solved, each requiring multiple restarts to circumvent local optima. In this work, we propose a machine learning enabled, surrogate-based framework that replaces the expensive adjoint solver, without compromising on predicting predictive accuracy. Specifically, we first train a deep neural network (DNN) from training data generated from evaluating the high-fidelity simulation model on a model-agnostic design of experiments on the geometry shape parameters. The optimum shape may then be computed by using a gradient-based optimizer coupled with the trained DNN. Subsequently, we also perform a gradient-free Bayesian optimization, where the trained DNN is used as the prior mean. We observe that the latter framework (DNN-BO) improves upon the DNN-only based optimization strategy for the same computational cost. Overall, this framework predicts the true optimum with very high accuracy, while requiring far fewer high-fidelity function calls compared to the adjoint-based method. Furthermore, we show that multiple optimization problems can be solved with the same machine learning model with high accuracy, to amortize the offline costs associated with constructing our models. Our methodology finds applications in the early stages of aerospace design. (C) 2021 Published by Elsevier Masson SAS.
Renganathan, S. Ashwin↗
Robust Acoustic Objective Functions and Sensitivities in Adjoint-Based Design Optimizations
The multidisciplinary design of aircraft typically includes considerations of performance, weight, fuel burn, and noise, among other factors. An objective function is applied to each of these considerations in order to weight the influence of trade-offs between different designs. Higher-order optimization exercises have utilized an adjoint approach to reach an optimal set of objective functions, such as maximum lift and reduced drag. Taking advantage of the adjoint approach significantly reduces the computational time required to find an optimal configuration. Including acoustics in the set of objective functions during an adjoint-based design optimization requires the sensitivity of the acoustic objective function. This document will present an approach for defining the sensitivity of several acoustic metrics and operations that can fill the role of the acoustic objective function. This includes time-integrated metrics such as effective perceived noise level (EPNL) and frequency-integrated metrics such as overall sound pressure level (OASPL). A demonstration case, validation, and details on the implementation in the second generation Aircraft NOise Prediction Program (ANOPP2) are also shown.
Lopes, Leonard V.↗
Discrete Adjoint-Based Design Optimization of Unsteady Turbulent Flows on Dynamic Unstructured Grids
An adjoint-based methodology for design optimization of unsteady turbulent flows on dynamic unstructured grids is described. The implementation relies on an existing unsteady three-dimensional unstructured grid solver capable of dynamic mesh simulations and discrete adjoint capabilities previously developed for steady flows. The discrete equations for the primal and adjoint systems are presented for the backward-difference family of time-integration schemes on both static and dynamic grids. The consistency of sensitivity derivatives is established via comparisons with complex-variable computations. The current work is believed to be the first verified implementation of an adjoint-based optimization methodology for the true time-dependent formulation of the Navier-Stokes equations in a practical computational code. Large-scale shape optimizations are demonstrated for turbulent flows over a tiltrotor geometry and a simulated aeroelastic motion of a fighter jet.
Nielsen, Eric J.↗
Constrained Multipoint Aerodynamic Shape Optimization Using an Adjoint Formulation and Parallel Computers
An aerodynamic shape optimization method that treats the design of complex aircraft configurations subject to high fidelity computational fluid dynamics (CFD), geometric constraints and multiple design points is described. The design process will be greatly accelerated through the use of both control theory and distributed memory computer architectures. Control theory is employed to derive the adjoint differential equations whose solution allows for the evaluation of design gradient information at a fraction of the computational cost required by previous design methods. The resulting problem is implemented on parallel distributed memory architectures using a domain decomposition approach, an optimized communication schedule, and the MPI (Message Passing Interface) standard for portability and efficiency. The final result achieves very rapid aerodynamic design based on a higher order CFD method. In order to facilitate the integration of these high fidelity CFD approaches into future multi-disciplinary optimization (NW) applications, new methods must be developed which are capable of simultaneously addressing complex geometries, multiple objective functions, and geometric design constraints. In our earlier studies, we coupled the adjoint based design formulations with unconstrained optimization algorithms and showed that the approach was effective for the aerodynamic design of airfoils, wings, wing-bodies, and complex aircraft configurations. In many of the results presented in these earlier works, geometric constraints were satisfied either by a projection into feasible space or by posing the design space parameterization such that it automatically satisfied constraints. Furthermore, with the exception of reference 9 where the second author initially explored the use of multipoint design in conjunction with adjoint formulations, our earlier works have focused on single point design efforts. Here we demonstrate that the same methodology may be extended to treat complete configuration designs subject to multiple design points and geometric constraints. Examples are presented for both transonic and supersonic configurations ranging from wing alone designs to complex configuration designs involving wing, fuselage, nacelles and pylons.
Reuther, James↗
Supersonic wing and wing-body shape optimization using an adjoint formulation
This paper describes the implementation of optimization techniques based on control theory for wing and wing-body design of supersonic configurations. The work represents an extension of our earlier research in which control theory is used to devise a design procedure that significantly reduces the computational cost by employing an adjoint equation. In previous studies it was shown that control theory could be used to~eviseransonic design methods for airfoils and wings in which the shape and the surrounding body-fitted mesh are both generated analytically, and the control is the mapping function. The method has also been implemented for both transonic potential flows and transonic flows governed by the Euler equations using an alternative formulation which employs numerically generated grids, so that it can treat more general configurations. Here results are presented for three-dimensional design cases subject to supersonic flows governed by the Euler equation.
Reuther, James↗
Overall Cooling Effectiveness With Internal Serpentine Channels and Optimized Film Cooling Holes
Abstract The overall cooling effectiveness for gas turbine airfoils is a function of the combined cooling due to internal cooling configurations and film cooling configurations. Typically, film cooling configurations are evaluated independent of the cooling effects of the internal feed channels, generally based on adiabatic effectiveness measurements. In this study, we consider the coupled effects of internal cooling and film cooling configurations through measurements of overall cooling effectiveness for film cooling holes fed by a coflow/counterflow channel and a serpentine channel. A film cooling hole designed by adjoint optimization techniques (X-AOpt) is compared to a standard-shaped hole with 7 deg forward and lateral expansions (7-7-7 SI). Experiments without film cooling showed that the serpentine channel had 35–50% greater overall cooling effectiveness than the straight, coflow channel. Experiments with the X-AOpt hole combined with a serpentine channel showed an area-averaged overall cooling effectiveness of ϕ¯¯=0.58, which was a 70% increase compared to the overall cooling effectiveness of the serpentine channel without film cooling. When the X-AOpt hole was fed with a coflow channel with similar coolant mass flowrate, the overall cooling effectiveness was ϕ¯¯=0.44, i.e., 30% lower than when using the serpentine channel. Interestingly, adiabatic effectiveness measurements with the X-AOpt holes showed a more uniform hole-to-hole performance when using the serpentine channel compared to the coflow channel.
Engineering↗
When ancient numerical demons meet physics-informed machine learning: adjoint-based gradients for implicit differentiable modeling
Recent advances in differentiable modeling, a genre of physics-informed machine learning that trains neural networks (NNs) together with process-based equations, have shown promise in enhancing hydrological models' accuracy, interpretability, and knowledge-discovery potential. Current differentiable models are efficient for NN-based parameter regionalization, but the simple explicit numerical schemes paired with sequential calculations (operator splitting) can incur numerical errors whose impacts on models' representation power and learned parameters are not clear. Implicit schemes, however, cannot rely on automatic differentiation to calculate gradients due to potential issues of gradient vanishing and memory demand. Here we propose a “discretize-then-optimize” adjoint method to enable differentiable implicit numerical schemes for the first time for large-scale hydrological modeling. The adjoint model demonstrates comprehensively improved performance, with Kling–Gupta efficiency coefficients, peak-flow and low-flow metrics, and evapotranspiration that moderately surpass the already-competitive explicit model. Therefore, the previous sequential-calculation approach had a detrimental impact on the model's ability to represent hydrological dynamics. Furthermore, with a structural update that describes capillary rise, the adjoint model can better describe baseflow in arid regions and also produce low flows that outperform even pure machine learning methods such as long short-term memory networks. The adjoint model rectified some parameter distortions but did not alter spatial parameter distributions, demonstrating the robustness of regionalized parameterization. Despite higher computational expenses and modest improvements, the adjoint model's success removes the barrier for complex implicit schemes to enrich differentiable modeling in hydrology.
58 GEOSCIENCES↗
Isolating Curvature Effects in Computing Wall-Bounded Turbulent Flows
The flow over the zero-pressure-gradient So-Mellor convex curved wall is simulated using the Navier-Stokes equations. An inviscid effective outer wall shape, undocumented in the experiment, is obtained by using an adjoint optimization method with the desired pressure distribution on the inner wall as the cost function. Using this wall shape with a Navier-Stokes method, the abilities of various turbulence models to simulate the effects of curvature without the complicating factor of streamwise pressure gradient can be evaluated. The one-equation Spalart-Allmaras turbulence model overpredicts eddy viscosity, and its boundary layer profiles are too full. A curvature-corrected version of this model improves results, which are sensitive to the choice of a particular constant. An explicit algebraic stress model does a reasonable job predicting this flow field. However, results can be slightly improved by modifying the assumption on anisotropy equilibrium in the model's derivation. The resulting curvature-corrected explicit algebraic stress model possesses no heuristic functions or additional constants. It lowers slightly the computed skin friction coefficient and the turbulent stress levels for this case (in better agreement with experiment), but the effect on computed velocity profiles is very small.
Rumsey, Christopher L.↗
Anisotropic Solution Adaptive Unstructured Grid Generation Using AFLR
An existing volume grid generation procedure, AFLR3, was successfully modified to generate anisotropic tetrahedral elements using a directional metric transformation defined at source nodes. The procedure can be coupled with a solver and an error estimator as part of an overall anisotropic solution adaptation methodology. It is suitable for use with an error estimator based on an adjoint, optimization, sensitivity derivative, or related approach. This offers many advantages, including more efficient point placement along with robust and efficient error estimation. It also serves as a framework for true grid optimization wherein error estimation and computational resources can be used as cost functions to determine the optimal point distribution. Within AFLR3 the metric transformation is implemented using a set of transformation vectors and associated aspect ratios. The modified overall procedure is presented along with details of the anisotropic transformation implementation. Multiple two-and three-dimensional examples are also presented that demonstrate the capability of the modified AFLR procedure to generate anisotropic elements using a set of source nodes with anisotropic transformation metrics. The example cases presented use moderate levels of anisotropy and result in usable element quality. Future testing with various flow solvers and methods for obtaining transformation metric information is needed to determine practical limits and evaluate the efficacy of the overall approach.
Marcum, David L.↗
Aerodynamic Design of the Hybrid Wing Body Propulsion-Airframe Integration
A hybrid wingbody (HWB) concept is being considered by NASA as a potential subsonic transport aircraft that meets aerodynamic, fuel, emission, and noise goals in the time frame of the 2030s. While the concept promises advantages over conventional wing-and-tube aircraft, it poses unknowns and risks, thus requiring in-depth and broad assessments. Specifically, the configuration entails a tight integration of the airframe and propulsion geometries; the aerodynamic impact has to be carefully evaluated. With the propulsion nacelle installed on the (upper) body, the lift and drag are affected by the mutual interference effects between the airframe and nacelle. The static margin for longitudinal stability is also adversely changed. We develop a design approach in which the integrated geometry of airframe (HWB) and propulsion is accounted for simultaneously in a simple algebraic manner, via parameterization of the planform and airfoils at control sections of the wingbody. In this paper, we present the design of a 300-passenger transport that employs distributed electric fans for propulsion. The trim for stability is achieved through the use of the wingtip twist angle. The geometric shape variables are determined through the adjoint optimization method by minimizing the drag while subject to lift, pitch moment, and geometry constraints. The design results clearly show the influence on the aerodynamic characteristics of the installed nacelle and trimming for stability. A drag minimization with the trim constraint yields a reduction of 10 counts in the drag coefficient.
hybrid wing↗
Comparison of MERRA-2 and ECCO-V4 Ocean Surface Heat Fluxes: Consequences of Different Forcing Feedbacks on Ocean Circulation and Implications for Climate Data Assimilation
Forcing ocean models with reanalysis data is a common practice in ocean modeling. As part of this practice, prescribed atmospheric state variables and interactive ocean SST (Sea Surface Temperature) are used to calculate fluxes between the ocean and the atmosphere. When forcing an ocean model with reanalysis fields, errors in the reanalysis data, errors in the ocean model and errors in the forcing formulation will generate a different solution compared to other ocean reanalysis solutions (which also have their own errors). As a first step towards a consistent coupled ocean-atmosphere reanalysis, we compare surface heat fluxes from a state-of-the-art atmospheric reanalysis, the Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2), to heat fluxes from a state-of-the-art oceanic reanalysis, the Estimating the Circulation and Climate of the Ocean Version 4, Release 2 (ECCO-v4). Then, we investigate the errors associated with the MITgcm (Massachusetts Institute of Technology general circulation model) ocean model in its ECCO-v4 ocean reanalysis configuration (1992-2011) when it is forced with MERRA- 2 atmospheric reanalysis fields instead of with the ECCO-v4 adjoint optimized ERA-interim state variables. This is done by forcing ECCO-v4 ocean with and without feedbacks from MERRA-2 related to turbulent fluxes of heat and moisture and the outgoing long wave radiation. In addition, we introduce an intermediate forcing method that includes only the feedback from the interactive outgoing long wave radiation. The resulting ocean circulation is compared with ECCO-v4 reanalysis and in-situ observations. We show that, without feedbacks, imbalances in the energy and the hydrological cycles of MERRA-2 (which are directly related to the fact it was created without interactive ocean) result in considerable SST drifts and a large reduction in sea level. The bulk formulae and interactive outgoing long wave radiation, although providing air-sea feedbacks and reducing model-data misfit, strongly relax the ocean to observed SST and may result in unwanted features such as large change in the water budget. These features have implications in a desired forcing recipe to be used. The results strongly and unambiguously argue for next generation data assimilation climate studies to involve fully coupled systems.
MERRA-2↗
Failure Analysis of the Method of Adjoint Systems to Determine Optimal Trajectories
Failure analysis of adjoint systems method for determining optimal trajectories
FAILURE MODE↗
Investigation of a Smooth Local Correlation-based Transition Model in a Discrete-Adjoint Aerodynamic Shape Optimization Algorithm
A smooth local correlation-based transition model is fully coupled to a RANS-based Newton-Krylov flow solver and discrete-adjoint gradient-based optimization algorithm. The free-transition optimization framework is evaluated using lift-constrained drag minimizations of airfoils at design conditions ranging from light to single-aisle aircraft and an infinite swept wing at design conditions representative of a transonic strut-braced wing aircraft. The impact of the streamwise grid resolution on the ability of the optimization algorithm to delay boundary-layer transition is investigated, with the results demonstrating that streamwise grid resolution requirements increase as the transition length decreases with increasing Reynolds number. The optimization problem at the light aircraft design conditions is demonstrated to be multi-modal, with the optimization algorithm producing two distinct designs: one with a thin, reflexed trailing edge and steep pressure recovery regions, the other with increased aft loading, with the latter design outperforming the former. A drag minimization of an airfoil at transonic design conditions demonstrates that the optimization algorithm successfully trades a decrease in viscous drag by delaying boundary-layer transition with an increase in wave drag, while the drag minimization of an infinite swept wing demonstrates the capability of the optimizational gorithm to delay both Tollmien-Schlichting and stationary crossflow instabilities.
AATT↗
Local-in-Time Adjoint-Based Method for Optimal Control/Design Optimization of Unsteady Compressible Flows
.We study local-in-time adjoint-based methods for minimization of ow matching functionals subject to the 2-D unsteady compressible Euler equations. The key idea of the local-in-time method is to construct a very accurate approximation of the global-in-time adjoint equations and the corresponding sensitivity derivative by using only local information available on each time subinterval. In contrast to conventional time-dependent adjoint-based optimization methods which require backward-in-time integration of the adjoint equations over the entire time interval, the local-in-time method solves local adjoint equations sequentially over each time subinterval. Since each subinterval contains relatively few time steps, the storage cost of the local-in-time method is much lower than that of the global adjoint formulation, thus making the time-dependent optimization feasible for practical applications. The paper presents a detailed comparison of the local- and global-in-time adjoint-based methods for minimization of a tracking functional governed by the Euler equations describing the ow around a circular bump. Our numerical results show that the local-in-time method converges to the same optimal solution obtained with the global counterpart, while drastically reducing the memory cost as compared to the global-in-time adjoint formulation.
Yamaleev, N. K.↗
Optimal checkpointing for adjoint multistage time-stepping schemes
Here, we consider checkpointing strategies that minimize the number of recomputations needed when performing discrete adjoint computations using multistage time-stepping schemes that require computing several substeps within one complete time step. Specifically, we propose two algorithms that can generate optimal checkpoint-ing schedules under weak assumptions. The first is an extension of the seminal Revolve algorithm adapted to multistage schemes. The second algorithm, named CAMS, is developed based on dynamic programming, and it requires the least number of recomputations when compared with other algorithms. The CAMS algorithm is made publicly available in a library with bindings to C and Python. Numerical results show that the proposed algorithms can deliver up to two times the speedup compared with that of classical Revolve. Moreover, we discuss the utilization of the CAMS library in mature scientific computing libraries and demonstrate the ease of using it in an adjoint workflow. The proposed algorithms have been adopted by the PETSc TSAdjoint library. Their performance has been demonstrated with a large-scale PDE-constrained optimization problem on a leadership-class supercomputer. This work is a significant extension of the authors' conference paper.
97 MATHEMATICS AND COMPUTING↗