SEARCH · Search NASA

Results for “projected gradient method”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 235 records · Page 13

All sources NASA NTRS DOE OSTI

DOE OSTI2022

Study of Self-Interaction Errors in Density Functional Calculations of Magnetic Exchange Coupling Constants Using Three Self-Interaction Correction Methods

We examine the role of self-interaction errors (SIE) removal on the evaluation of magnetic exchange coupling constants. In particular we analyze the effect of scaling down the self-interaction-correction (SIC) for three non-empirical density functional approximations (DFAs) namely, the local spin density approximation, the Perdew-Burke-Ernzerhof generalized gradient approximation, and recent SCAN family of meta-GGA functionals. To this end, we employ three one-electron SIC methods: Perdew-Zunger [Perdew, J. P.; Zunger, A. Phys. Rev. B, 1981, 23, 5048] SIC, the orbitalwise scaled SIC method [Vydrov, O. A. et al., J. Chem. Phys. 2006, 124, 094108], and the recent local scaling method [Zope, R. R. et al., J. Chem. Phys. 2019, 151, 214108]. We compute the magnetic exchange coupling constants using the spin projection and non projection approaches for sets of molecules composed of dinuclear and polynuclear H-He models, organic radical molecules, and chlorocuprate, and compare these results against accurate theories and experiment. Our results show that for the systems that mainly consist of single electron regions, PZSIC performs well but for more complex organic systems and the chlorcuprates, an overcorrecting tendency of PZSIC combined with the DFAs utilized in this work is more pronounced, and in such cases LSIC with kinetic energy density ratio performs better than PZSIC. Furthermore, analysis of the results in terms of SIC corrections to the density and to the total energy shows that both density and energy correction are required to obtain an improved prediction of magnetic exchange couplings.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Trustworthy Physics-Informed Deep Learning for Predictive Scientific Computing

This project has developed powerful trustworthy physics-informed deep learning (TPiDL) models and methods to fundamentally enhance the scale and power of computational modeling in the scientific and engineering domains. Deep learning (DL) has radically advanced the state-of-the-art in machine learning, computer vision, natural language processing, and also scientific computing. Nevertheless, progress has been driven almost entirely by empirical observations, hacks, and tricks. Under the support of this project, the graph operator learning tools and advanced trustworthy physical informed neural networks have been developed. In addition, stochastic gradient replica-exchange Markov Chain Monte Carlo (MCMC) sampling algorithms have been designed to quantify the uncertainties and speed up the training of large-scale neural networks.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

MAPPRAISER: A massively parallel map-making framework for multi-kilo pixel CMB experiments

Forthcoming cosmic microwave background (CMB) polarized anisotropy experiments have the potential to revolutionize our understanding of the Universe and fundamental physics. The sought-after, tale-telling signatures will be however distributed over voluminous data sets which these experiments will collect. These data sets will need to be efficiently processed and unwanted contributions due to astrophysical, environmental, and instrumental effects characterized and efficiently mitigated in order to uncover the signatures. This poses a significant challenge to data analysis methods, techniques, and software tools which will not only have to be able to cope with huge volumes of data but to do so with unprecedented precision driven by the demanding science goals posed for the new experiments. A keystone of efficient CMB data analysis is solvers of very large linear systems of equations. Such systems appear in very diverse contexts throughout CMB data analysis pipelines, however they typically display similar algebraic structures and can therefore be solved using similar numerical techniques. Linear systems arising in the so-called map-making problem are one of the most prominent and common ones. In this work we present a massively parallel, flexible and extensible framework, comprised of a numerical library, MIDAPACK, and a high level code, MAPPRAISER, which provide tools for solving efficiently such systems. Here, the framework implements iterative solvers based on conjugate gradient techniques: enlarged and preconditioned using different preconditioners. We demonstrate the framework on simulated examples reflecting basic characteristics of the forthcoming data sets issued by ground-based and satellite-borne instruments, executing it on as many as 16,384 compute cores. The software is developed as an open source project freely available to the community at: https://github.com/B3Dcmb/midapack.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1973

Theoretical autophotogrammetry. I - The method of the photometric potential

The present work presents preliminary efforts at a mathematical formulation of a photogrammetric method suitable for computerized processing. A theory is derived for the extraction of surface topography from stereometric pairs of photographs. The theory can be applied to either convergent or near-orthographic-projection stereo. Scene brightness is treated as a scalar function defined on a general surface embedded in ordinary three-dimensional space (the actual topography). Correspondence of points is based on adjustments leading to photometric equivalence, while the adjustments are based on the local value of the photometric gradient.

Wildey, R. L.↗

DOE OSTI2024

Ultra-compact and efficient heat exchanger for solar desalination with unprecedented scaling resistance

The objective is to design, prototype, and test a heat exchanger technology with an efficiency improvement of more than 150% compared to state-of-the-art heat exchanger technology (UA/volume), with significantly improved fouling/scaling resistance at elevated temperatures (>120°C) when compared to existing technology where scaling is a major performance-limiting issue, as well as at lower operating-temperatures (~63-110°C). We now report our overall progress on the project. An ultra-compact, efficient heat exchanger geometry has been developed that can maximize surface-area-to-volume ratio so that proposed target (UA/volume). Advanced manufacturing techniques, such additive manufacturing, have made it possible to create complex structures cost effectively. In addition to traditional manufacturing methods, additive manufacturing has been successfully used for the development of high-performance geometries. These methods allow the ability to create custom surface textures, which are deployed to dramatically control wettability and create energy-gradient surfaces. These techniques allow fouling/scaling nucleation to be spatially controlled, such that as the fouling layer grows it will be broken and removed by flow forces.

42 ENGINEERING↗

NASA NTRS2006

Refinement of Methods for Evaluation of Near-Hypersingular Integrals in BEM Formulations

In this paper, we present advances in singularity cancellation techniques applied to integrals in BEM formulations that are nearly hypersingular. Significant advances have been made recently in singularity cancellation techniques applied to 1 R type kernels [M. Khayat, D. Wilton, IEEE Trans. Antennas and Prop., 53, pp. 3180-3190, 2005], as well as to the gradients of these kernels [P. Fink, D. Wilton, and M. Khayat, Proc. ICEAA, pp. 861-864, Torino, Italy, 2005] on curved subdomains. In these approaches, the source triangle is divided into three tangent subtriangles with a common vertex at the normal projection of the observation point onto the source element or the extended surface containing it. The geometry of a typical tangent subtriangle and its local rectangular coordinate system with origin at the projected observation point is shown in Fig. 1. Whereas singularity cancellation techniques for 1 R type kernels are now nearing maturity, the efficient handling of near-hypersingular kernels still needs attention. For example, in the gradient reference above, techniques are presented for computing the normal component of the gradient relative to the plane containing the tangent subtriangle. These techniques, summarized in the transformations in Table 1, are applied at the sub-triangle level and correspond particularly to the case in which the normal projection of the observation point lies within the boundary of the source element. They are found to be highly efficient as z approaches zero. Here, we extend the approach to cover two instances not previously addressed. First, we consider the case in which the normal projection of the observation point lies external to the source element. For such cases, we find that simple modifications to the transformations of Table 1 permit significant savings in computational cost. Second, we present techniques that permit accurate computation of the tangential components of the gradient; i.e., tangent to the plane containing the source element.

Fink, Patricia W.↗

DOE OSTI2024

Distributed optimization for multi-commodity urban traffic control

A distributed method for concurrent traffic signal and routing control of traffic networks is proposed. The method is based on the multi-commodity store-and-forward model, in which the destinations are the commodities. The system benefits from the communication between vehicles and infrastructure, providing optimal signal timings to intersections and routes to vehicles on a link-by-link basis. Using the augmented Lagrangian to model the constraints into the objective, the baseline centralized problem is decomposed into a set of objective-coupled subproblems, one for each intersection, enabling the solution to be computed by a distributed- gradient projection algorithm. Further, the intersection agents only need to communicate and coordinate with neighboring intersections to ensure convergence to the optimal solution while tolerating suboptimal iterations that offer more flexibility, unlike other distributed approaches. Through microsimulation, we demonstrate the effectiveness of the proposed algorithm in traffic networks with time-varying demand. Computational analysis shows that the distributed problem is suitable for real-time applications. A robustness analysis show that the distributed formulation enables a graceful degradation of the system in case of failure.

Augmented Lagrangian↗

DOE OSTI2016

A Numerical Method for High Energy X-Ray Source Shape Reconstruction

Quantifying the source shape of a pulsed power X-ray imaging system is essential in determining a lower bound for the overall resolution of the system. A method of reconstructing the source shape involves solving an inverse problem with an operator matrix of size n x n, where n is the number of pixels in the image. Building and inverting such a large matrix quickly becomes intractable as n increases due to both memory and computational constraints, even in well-posed problems. The ill-posed nature of this problem adds further complications, making direct inversion unstable. Taking advantage of the structure of the operator matrix and utilizing the Projected Newton and Conjugate Gradient algorithms, we are able to perform reconstructions of images that are orders of magnitude larger than previously possible.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS2003

Citizen Explorer: An Earth Observer With New Small Satellite Technology - 1

Citizen Explorer-I (CX-I), designed and built by students at Colorado Space Grant Consortium in Boulder to provide global ozone monitoring, employs a unique mission architecture and several innovative technologies during its mission. The mission design allows K-12 schools around the world to be involved as ground stations available to receive science data and telemetry from CX-I. Another important technology allows the spacecraft to be less reliant on ground operators. Spacecraft Command Language (SCL) allows mission designers to set constraints on the satellite operations. The satellite then automatically adheres to the constraints when the satellite is out of contact with Mission Operations. In addition to SCL, a low level of artificial intelligence will be supplied to the spacecraft through the use of the Automated Scheduling and Planning ENvironment (ASPEN). ASPEN is used to maintain a spacecraft schedule in order to achieve the objectives a mission operator would normally have to complete. Within the communications system of CX-I, internet of CX-I, internet protocols are the main method for communicating with the satellite. As internet protocols have not been widely used in satellite communication, CX-I provides an opportunity to study the effectiveness of using internet protocols over radio links. The Attitude Determination and Control System (ADCS) on CX-I uses a gravity gradient boom as a means of orienting the satellite's science instruments toward nadir. The boom design is unique because it is constructed of tape measure material. These new technologies' effectiveness will be tested for use on future small satellite projects within the space satellite industry.

Allen, Zachary↗

DOE OSTI2016

SU-F-T-316: A Model to Deal with Dosimetric and Delivery Uncertainties in Radiotherapy Treatment Planning

Purpose The conventional way of dealing with uncertainties resulting from dose calculation or beam delivery in IMRT, is to do verification measurements for the plan in question. Here we present an alternative based on recommendations given in the AAPM 142 report and treatment specific parameters that model the uncertainties for the plan delivery. Methods Basis of the model is the assignment of uncertainty parameters to all segment fields or control point sequences of a plan. The given field shape is analyzed for complexity, dose rate, number of MU, field size related output as well as factors for in/out field position and penumbra regions. Together with depth related uncertainties, a 3D matrix is generated by a projection algorithm. Patient anatomy is included as uncertainty CT data set as well. Therefore, object density is classified in 4 categories close to water, lung, bone and gradient regions with additional uncertainties. The result is then exported as a DICOM dose file by the software tool (written in IDL, Exelis), having the given resolution and target point. Results Uncertainty matrixes for several patient cases have been calculated and compared side by side in the planning system. The result is not quite always intuitive but it clearly indicates high and low uncertainties related to OARs and target volumes as well as to measured gamma distributions.ConclusionThe imported uncertainty datasets may help the treatment planner to understand the complexity of the treatment plan. He then might decide to change the plan to produce a more suited uncertainty distribution, e.g. by changing the beam angles the high uncertainty spots can be influenced or try to use another treatment setup, resulting in a plan with lower uncertainties. A next step could be to include such a model into the optimization algorithm to add a new dose uncertainty constraint.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2019

The LCLS-II HE High Q and Gradient R&D Program

The LCLS-II HE project is a high energy upgrade to the superconducting LCLS-II linac. It consists of adding twenty additional 1.3 GHz cryomodules to the linac, with cavities operating at a gradient of 20.8 MV/m with a Q₀ of 2.7·10¹⁰. Performance of LCLS-II cryomodules has suggested that operations at this high of a gradient will not be achievable with the existing cavity recipe employed. Therefore a research program was developed between SLAC, Fermilab, Thomas Jefferson National Accelerator Facility, and Cornell University in order to improve the cavity processing method of the SRF cavities and reach the HE goals. This program explores the doping regime beyond what was done for LCLS-II and also has looked to further developed nitrogen-infusion. Here we will summarize the results from this R\&D program, showing significant improvement on both single-cell and 9-cell cavities compared with the original LCLS-II cavity recipe.

43 PARTICLE ACCELERATORS↗

DOE OSTI2020

A Look At Dissolved Organic Carbon In Streams and Its Effect On DGT Hg Collection

Savannah River Site (SRS) and Oak Ridge National Laboratory (ORNL) were both major material production sites for the U.S.'s nuclear program during the 20. century. Both sites discharged mercury (Hg) to streams. Oak Ridge streams have accumulated significant levels of Hg from early 'historical' discharges. Diffusive Gradient in Thin Films (DGT) samplers may be a labor and material saving method to monitor the Hg contamination of streams at both sites. Humic and Fulvic acids are released during the degradation of plant and animal residue. They are part of the Dissolved Organic Matter that colors the black water streams of the southern U.S.. Humic acid and Fulvic acids have a tendency to sequester heavy metals, like Hg, and may interfere with the reactions needed to draw the Hg through the thin films of the DGT samplers. Project Objective: Use a UV-Vis spectrophotometer to estimate the levels of Dissolved Organic Carbon (DOC) in the streams selected to test the effect of DOC on DGT samplers. Conclusion: UV-Vis is a convenient way to estimate the amount of DOC in streams and may be used to provide more context in DGT Hg sampling. efforts. Future Work: Analyze data from SRS and ORNL DGT samples; Compare the results of the DGT samplers with more traditional speciating methods from ORNL; Use a TOC Spectrophotometer to gain a more accurate picture of stream DOC.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2020

MimicGAN: Robust Projection onto Image Manifolds with Corruption Mimicking

In the past few years, Generative Adversarial Networks (GANs) have dramatically advanced our ability to represent and parameterize high-dimensional, non-linear image manifolds. As a result, they have been widely adopted across a variety of applications, ranging from challenging inverse problems like image completion, to problems such as anomaly detection and adversarial defense. A recurring theme in many of these applications is the notion of projecting an image observation onto the manifold that is inferred by the generator. In this context, Projected Gradient Descent (PGD) has been the most popular approach, which essentially optimizes for a latent vector that minimizes the discrepancy between a generated image and the given observation. However, PGD is a brittle optimization technique that fails to identify the right projection (or latent vector) when the observation is corrupted, or perturbed even by a small amount. Such corruptions are common in the real world, for example images in the wild come with unknown crops, rotations, missing pixels, or other kinds of non-linear distributional shifts which break current encoding methods, rendering downstream applications unusable. To address this, we propose corruption mimicking—a new robust projection technique, that utilizes a surrogate network to approximate the unknown corruption directly at test time, without the need for additional supervision or data augmentation. The proposed method is significantly more robust than PGD and other competing methods under a wide variety of corruptions, thereby enabling a more effective use of GANs in real-world applications. Finally, more importantly, we show that our approach produces state-of-the-art performance in several GAN-based applications—anomaly detection, domain adaptation, and adversarial defense, that benefit from an accurate projection.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Lensing Constraints on the Mass Profile Shape and the Splashback Radius of Galaxy Clusters

The lensing signal around galaxy clusters can, in principle, be used to test detailed predictions for their average mass profile from numerical simulations. However, the intrinsic shape of the profiles can be smeared out when a sample that spans a wide range of cluster masses is averaged in physical length units. This effect especially conceals rapid changes in gradient such as the steep drop associated with the splashback radius, a sharp edge corresponding to the outermost caustic in accreting halos. We optimize the extraction of such local features by scaling individual halo profiles to a number of spherical overdensity radii, and apply this method to 16 X-ray-selected, high-mass clusters targeted in the Cluster Lensing And Supernova survey with Hubble. By forward-modeling the weak- and strong-lensing data presented by Umetsu et al., we show that, regardless of the scaling overdensity, the projected ensemble density profile is remarkably well described by a Navarro–Frenk–White (NFW) or Einasto profile out to R∼2.5 h{sup −1} Mpc, beyond which the profiles flatten. We constrain the NFW concentration to c{sub 200c}=3.66±0.11 at M{sub 200c}≃1.0×10{sup 15} h{sup −1} M{sub ⊙}, consistent with and improved from previous work that used conventionally stacked lensing profiles, and in excellent agreement with theoretical expectations. Assuming the profile form of Diemer and Kravtsov and generic priors calibrated from numerical simulations, we place a lower limit on the splashback radius of the cluster halos, if it exists, of R{sub sp}{sup 3D}/r{sub 200m}>0.89 (R{sub sp}{sup 3D}>1.83 h{sup −1} Mpc) at 68% confidence. The corresponding density feature is most pronounced when the cluster profiles are scaled by r{sub 200m}, and smeared out when scaled to higher overdensities.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2016

Final Technical Report: Using Solid Particles as Heat Transfer Fluid for use in Concentrating Solar Power (CSP) Plants

In today’s industrial economy, energy consumption has never been higher. Over the last 15 years the US alone has consumed an average of nearly 100 quadrillion BTUs per year [21]. A need for clean and renewable energy sources has become quite apparent. The SunShot Initiative is an ambitious effort taken on by the United States Department of Energy that targets the development of solar energy that is cost-competitive with other methods for generating electricity. Specifically, this work is concerned with the development of concentrating solar power plants (CSPs) with granular media as the heat transfer fluid (HTF) from the solar receiver. Unfortunately, the prediction of heat transfer in multiphase flows is not well understood. For this reason, our aim is to fundamentally advance the understanding of multiphase heat transfer, particularly in gas-solid flows, while providing quantitative input for the design of a near black body receiver (NBB) that uses solid grains (like sand) as the HTF. Over the course of this three-year project, a wide variety of contributions have been made to advance the state-of-the art description for non-radiative heat transfer in dense, gas-solid systems. Comparisons between a state-of-the-art continuum heat transfer model and discrete element method (DEM) simulations have been drawn. The results of these comparisons brought to light the limitations of the continuum model due to inherent assumptions in its derivation. A new continuum model was then developed for heat transfer at a solid boundary by rigorously accounting for the most dominant non-radiative heat transfer mechanism (particle-fluid-wall conduction). The new model is shown to be in excellent agreement with DEM data and captures the dependence of heat transfer on particle size, a dependency that previous continuum models were not capable of. DEM and the new continuum model were then employed to model heat transfer in a variety of receiver geometries. The results provided crucial feedback on the efficiency and feasibility of various designs. Namely, a prototype design consisting of an array of heated hexagonal tubes was later supplanted by a vertical conduit with internal baffles. Due to low solids heat transfer on the bottom faces of the hexagonal tubes in the prototype, the predicted wall temperature gradients exceeded the design limitations. By contrast, the vertical conduit can be constructed to continually force particle-wall contacts, and thus, result in more desirable solids heat transfer and wall temperature gradients. Finally, a new heat flux boundary condition was developed for DEM simulations to assess the aforementioned wall temperature gradients. The new boundary condition advances current state-of-the-art techniques by allowing the heat fluxes to each phase to vary with space and time while the total flux remains constant. Simulations with the new boundary condition show that the total boundary heat flux is in good agreement with the imposed total boundary heat flux. While the methods we have utilized here are primarily numerical and fundamental by nature, they offer some key advantages of: (i) being robust and valid over a large range of conditions, (ii) able to quickly explore large parameter spaces, and (iii) aid in the construction of experiments. We have ultimately leveraged our computational capabilities to provide feedback on the design of a CSP which possesses great potential to become a cost effective source of clean and renewable electricity. Overall, ensuring that future energy demands are met in a responsible and efficient manner has far reaching impacts that span both ecologic and economic concerns. Regarding logistics, the project was successfully re-negotiated after the go/no-decisions of Years 1 and 2. All milestones were successfully completed.

14 SOLAR ENERGY↗

DOE OSTI2020

Methods for Explainable Artificial Intelligence

We explored ways of quantifying information in a neural network. This can be used to determine the right size of a network or to infer the way in which a network is processing information. The first year and a half was somewhat exploratory while the last half of the project focused on approaches that seemed to show the most promise. The introduce a new way of computing explainable artificial intelligence (XAI) saliency maps that is several orders of magnitude faster than methods with similar fidelity We call it FastCAM. The method works be combining a Class Activation Map (CAM) method such as GradCAM with a forward activation map computed with a statistic we call SMOE Scale. The addition of the forward activation maps to CAM methods seems to always improve their fidelity. At the same time, computational overhead is not increased by very much. While Gradients with SmoothGrad scores better on some fidelity measures, it is overall not as good and requires more than 1500 times to compute. We demonstrate two completed applications of FastCAM on tasks outside of the LDRD at LLNL. The source code for FastCAM is currently being implanted into Captum, the official XAI toolkit for the popular deep learning toolkit PyTorch. The LDRD currently has 13 publications released to the public. 10 of them are journal length.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2016

Predicting Ice Sheet and Climate Evolution at Extreme Scales

A main research objectives of PISCEES is the development of formal methods for quantifying uncertainties in ice sheet modeling. Uncertainties in simulating and projecting mass loss from the polar ice sheets arise primarily from initial conditions, surface and basal boundary conditions, and model parameters. In general terms, two main chains of uncertainty propagation may be identified: 1. inverse propagation of observation and/or prior onto posterior control variable uncertainties; 2. forward propagation of prior or posterior control variable uncertainties onto those of target output quantities of interest (e.g., climate indices or ice sheet mass loss). A related goal is the development of computationally efficient methods for producing initial conditions for an ice sheet that are close to available present-day observations and essentially free of artificial model drift, which is required in order to be useful for model projections (“initialization problem”). To be of maximum value, such optimal initial states should be accompanied by “useful” uncertainty estimates that account for the different sources of uncerainties, as well as the degree to which the optimum state is constrained by available observations. The PISCEES proposal outlined two approaches for quantifying uncertainties. The first targets the full exploration of the uncertainty in model projections with sampling-based methods and a workflow managed by DAKOTA (the main delivery vehicle for software developed under QUEST). This is feasible for low-dimensional problems, e.g., those with a handful of global parameters to be inferred. This approach can benefit from derivative/adjoint information, but it is not necessary, which is why it often referred to as “non-intrusive”. The second approach makes heavy use of derivative information from model adjoints to address quantifying uncertainty in high-dimensions (e.g., basal boundary conditions in ice sheet models). The use of local gradient, or Hessian information (i.e., second derivatives of the cost function), requires additional code development and implementation, and is thus often referred to as an “intrusive” approach. Within PISCEES, MIT has been tasked to develop methods for derivative-based UQ, the ”intrusive” approach discussed above. These methods rely on the availability of first (adjoint) and second (Hessian) derivative code, developed through intrusive methods such as algorithmic differentiation (AD). While representing a significant burden in terms of code development, derivative-baesd UQ is able to cope with very high-dimensional uncertainty spaces. That is, unlike sampling methods (all variations of Monte Carlo), calculational burden is independent of the dimension of the uncertainty space. This is a significant advantage for spatially distributed uncertainty fields, such as threedimensional initial conditions, three-dimensional parameter fields, or two-dimensional surface and basal boundary conditions. Importantly, uncertainty fields for ice sheet models generally fall into this category.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Minimum feature size control in level set topology optimization via density fields

A level set topology optimization approach that uses an auxiliary density field to nucleate holes during the optimization process and achieves minimum feature size control in optimized designs is explored. The level set field determines the solid-void interface and the density field describes the distribution of a fictitious porous material using the solid isotropic material with penalization. These fields are governed by two sets of independent optimization variables which are initially coupled using a penalty for hole nucleation. The strength of the density field penalization and projection is gradually increased during the optimization process to promote a 0-1 density distribution. In addition, a second penalty regulates the evolution of the density field in the void phase. The treatment of the density field combined with the second penalty mitigate the appearance of small design features. The minimum feature size of optimized designs is controlled by the radius of the linear filter applied to the density optimization variables. The structural response is predicted by the extended finite element method, the sensitivities by the adjoint method, and the optimization variables are updated by a gradient-based optimization algorithm. Numerical examples investigate the robustness of this approach with respect to algorithmic parameters and mesh refinement. The results show the applicability of the combined density level set topology optimization approach for both optimal hole nucleation and for minimum feature size control in 2D and 3D. This comes, however, at the cost of a more complex problem formulation and additional computational cost due to an increased number of optimization variables.

42 ENGINEERING↗