Discrete Adjoint-Based Design Optimization of Unsteady Turbulent Flows on Dynamic Unstructured Grids

Nielsen, Eric J.; Diskin, Boris; Yamaleev, Nail K.

doi:10.2514/6.2009-3802

Cited by 24 publications

(28 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this manner, the file pointer can be placed immediately at the location of interest and I/O can proceed. This approach worked well for the cases shown in [28] using 1,024 processors. Those simulations utilized a Lustre-based parallel file system [14]; all results described here rely on identical hardware.…”

Section: Parallel Scaling and I/o Performancementioning

confidence: 87%

“…The implementation of the time-dependent adjoint and sensitivity equations has been previously described and verified in [28]. The general methodology for solving the adjoint equations is similar in nature and reverse in order to that used for the primal equations.…”

Section: Flapmentioning

confidence: 99%

“…Great care must be taken to perform these operations in a scalable manner as will be discussed below. It should also be noted that any issues related to disk I/O revealed in the current study would likely be magnified in the dynamic grid context, where the mesh coordinates and velocities are also stored at each time step [28]. Whether a single plane of steady data or several back planes of an unsteady solution are required for restart purposes, the solvers traditionally use a strategy in which every processor writes to a single file.…”

Section: Parallel Scaling and I/o Performancementioning

confidence: 99%

“…However, this approach becomes untenable for adjoint-based design, where solution data must be stored or loaded at every time step. As described in [28], the baseline strategy for meeting this more stringent requirement relies on each processor simultaneously accessing its own direct-access file, where each record of the file contains all of the data relevant to that partition for time level . In this manner, the file pointer can be placed immediately at the location of interest and I/O can proceed.…”

Section: Parallel Scaling and I/o Performancementioning

confidence: 99%

“…For the unsteady implementation described in [28], similar general cost functions are defined at each time level . The integrated cost function is defined as a discrete time integral over a certain time interval [ 1 , 2 ]:…”

Section: Cost Functions and Adjoint Equationsmentioning

confidence: 99%

See 4 more Smart Citations

Integrated Design of an Active Flow Control System Using a Time-Dependent Adjoint Method

Nielsen

Jones

2011

Math. Model. Nat. Phenom.

View full text Add to dashboard Cite

Abstract. An exploratory study is performed to investigate the use of a time-dependent discrete adjoint methodology for design optimization of a high-lift wing configuration augmented with an active flow control system. The location and blowing parameters associated with a series of jet actuation orifices are used as design variables. In addition, a geometric parameterization scheme is developed to provide a compact set of design variables describing the wing shape. The scaling of the implementation is studied using several thousand processors and it is found that asynchronous file operations can greatly improve the overall performance of the approach in such massively parallel environments. Three design examples are presented which seek to maximize the mean value of the lift coefficient for the coupled system, and results demonstrate improvements as high as 27% relative to the lift obtained with non-optimized actuation. This lift gain is more than three times the incremental lift provided by the non-optimized actuation.

show abstract

Section: Parallel Scaling and I/o Performancementioning

confidence: 87%

Section: Flapmentioning

confidence: 99%

Section: Parallel Scaling and I/o Performancementioning

confidence: 99%

Section: Parallel Scaling and I/o Performancementioning

confidence: 99%

Section: Cost Functions and Adjoint Equationsmentioning

confidence: 99%

See 3 more Smart Citations

Integrated Design of an Active Flow Control System Using a Time-Dependent Adjoint Method

Nielsen

Jones

2011

Math. Model. Nat. Phenom.

View full text Add to dashboard Cite

show abstract

Control of hydrothermal waves in a thermocapillary flow using a gradient‐based control strategy

Muldoon

2012

Numerical Methods in Fluids

View full text Add to dashboard Cite

SUMMARYThe problem of controlling the hydrothermal waves in a thermocapillary flow is addressed using a gradient‐based control strategy. The state equations are the two‐dimensional unsteady incompressible Navier–Stokes and energy equations under the Boussinesq approximation. The modeled problem is the ‘open boat’ process of crystal growth, the flow which is driven by Marangoni and buoyancy effects. The control is a spatially and temporally varying heat flux boundary condition at the free surface. The control that minimizes the hydrothermal waves is found using a conjugate gradient method, where the gradient of the objective function with respect to the control variables is obtained from solving a set of adjoint equations. The effectiveness of choices of the parameters governing the control algorithm is examined. Almost complete suppression of the hydrothermal waves is obtained for certain choices of the parameters governing the control algorithm. The numerical issues involved with finding the control using the optimizer are discussed, and the features of the resulting control are analyzed with the goal of understanding how it affects the flow.Copyright © 2012 John Wiley & Sons, Ltd.

show abstract

Partitioned solution of an unsteady adjoint for strongly coupled fluid-structure interactions and application to parameter identification of a one-dimensional problem

Degroote

Hojjat

Stavropoulou

et al. 2012

Struct Multidisc Optim

View full text Add to dashboard Cite

Unsteady fluid-structure interaction (FSI) simulations are generally time-consuming. Gradient-based methods are preferred to minimise the computational cost of parameter identification studies (and more in general optimisation) with a high number of parameters. However, calculating the cost function's gradient using finite differences becomes prohibitively expensive for a high number of parameters. Therefore, the adjoint equations of the unsteady FSI problem are solved to obtain this gradient at a cost almost independent of the number of parameters.Here, both the forward and the adjoint problems are solved in a partitioned way, which means that the flow equations and the structural equations are solved separately. The application of interest is the identification of the arterial wall's stiffness by comparing the motion of the arterial wall with a reference, possibly obtained from non-invasive imaging. Due to the strong interaction between the fluid and the structure, quasi-Newton coupling iterations are applied to stabilise the partitioned solution of both the forward and the adjoint problem.

show abstract

Discrete Adjoint-Based Design Optimization of Unsteady Turbulent Flows on Dynamic Unstructured Grids

Cited by 24 publications

References 41 publications

Integrated Design of an Active Flow Control System Using a Time-Dependent Adjoint Method

Integrated Design of an Active Flow Control System Using a Time-Dependent Adjoint Method

Control of hydrothermal waves in a thermocapillary flow using a gradient‐based control strategy

Partitioned solution of an unsteady adjoint for strongly coupled fluid-structure interactions and application to parameter identification of a one-dimensional problem

Contact Info

Product

Resources

About