Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Optimize”

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.

At least 775 records · Page 43

A flexible and differentiable coil proxy for stellarator equilibrium optimization

Balancing plasma performance and coil cost is a significant challenge when designing a stellarator power plant. Most current stellarator designs are produced through two-stage optimization: stage-1 for the equilibrium and stage-2 for a coil design that reproduces its magnetic configuration. Because few proxies connect both stages, two-stage optimization can produce plasmas that have high-quality physical properties but overly complex coils. In recent years, single-stage optimization has increasingly been used to optimize the plasma and coils simultaneously in order to improve the plasma–coil balance. However, all existing single-stage tools are specialized for filament coils, cannot model coil systems containing permanent magnets or dipole arrays, and continue to be challenged by numerical problems. The quasi-single-stage optimization finds a middle ground by integrating a coil optimization subproblem into stage-1 optimization. We present a flexible, differentiable coil complexity proxy based on the newly developed QUADCOIL coil optimization code. QUADCOIL is fast and can target realistic coil metrics and constraints that are unavailable to codes with comparable speed. We demonstrate the effectiveness and flexibility of the QUADCOIL proxy by presenting two QSS optimization studies. The first study produces a permanent magnet solution for the MUSE stellarator with 34% fewer magnets than previous solutions. The second study produces a coil solution for the ARIES-CS stellarator with 30% reductions in both peak and root-mean-square force.

Fu, Lanke [Princeton Plasma Physics Laboratory (PP↗

Bayesian optimization algorithms for accelerator physics

Accelerator physics relies on numerical algorithms to solve optimization problems in online accelerator control and tasks such as experimental design and model calibration in simulations. The effectiveness of optimization algorithms in discovering ideal solutions for complex challenges with limited resources often determines the problem complexity these methods can address. The accelerator physics community has recognized the advantages of Bayesian optimization algorithms, which leverage statistical surrogate models of objective functions to effectively address complex optimization challenges, especially in the presence of noise during accelerator operation and in resource-intensive physics simulations. In this review article, we offer a conceptual overview of applying Bayesian optimization techniques toward solving optimization problems in accelerator physics. We begin by providing a straightforward explanation of the essential components that make up Bayesian optimization techniques. We then give an overview of current and previous work applying and modifying these techniques to solve accelerator physics challenges. Finally, we explore practical implementation strategies for Bayesian optimization algorithms to maximize their performance, enabling users to effectively address complex optimization challenges in real-time beam control and accelerator design. Published by the American Physical Society 2024

43 PARTICLE ACCELERATORS↗

PABO: Pseudo Agent-Based Multi-Objective Bayesian Hyperparameter Optimization for Efficient Neural Accelerator Design

The ever increasing computational cost of Deep Neural Networks (DNN) and the demand for energy efficient hardware for DNN acceleration has made accuracy and hardware cost co-optimization for DNNs tremendously important, especially for edge devices. Owing to the large parameter space and cost of evaluating each parameter in the search space, manually tuning of DNN hyperparameters is impractical. Automatic joint DNN and hardware hyperparameter optimization is indispensable for such problems. Bayesian optimization-based approaches have shown promising results for hyperparameter optimization of DNNs. However, most of these techniques have been developed without considering the underlying hardware, thereby leading to inefficient designs. Further, the few works that perform joint optimization are not generalizable and mainly focus on CMOS-based architectures. In this work, we present a novel pseudo agent-based multiobjective hyperparameter optimization (PABO) for maximizing the DNN performance while obtaining low hardware cost. Compared to the existing methods, our work poses a theoretically different approach for joint optimization of accuracy and hardware cost and focuses on memristive crossbar based accelerators. PABO uses a supervisor agent to establish connections between the posterior Gaussian distribution models of network accuracy and hardware cost requirements. The agent reduces the mathematical complexity of the co-optimization problem by removing unnecessary computations and updates of acquisition functions, thereby achieving significant speed-ups for the optimization procedure. PABO outputs a Pareto frontier that underscores the trade-offs between designing high-accuracy and hardware efficiency. Our results demonstrate a superior performance compared to the state-of-the-art methods both in terms of accuracy and computational speed (~100x speed up).

Parsa, Maryam↗

Co-Optimization of Velocity and Charge-Depletion for Plug-In Hybrid Electric Vehicles: Accounting for Acceleration and Jerk Constraints

Abstract Recent advances in vehicle connectivity and automation technologies promote advanced control algorithms that co-optimize the longitudinal dynamics and powertrain operation of hybrid electric vehicles. Typically, a sequential optimization with the vehicle dynamics optimized followed by powertrain optimization is adopted to manage a number of complexities such as the inherent mixed-integer nature of the hybrid powertrain, the numerous state and control variables, the differing time scales of vehicle and powertrain subsystems, time-varying state constraints, and large horizon lengths. Instead, we solve the offline optimization problem in a centralize manner assuming exact knowledge of the lead vehicle's position over the entire trip by applying a discrete-time single shooting-based numerical approach, Discrete Mixed-Integer Shooting (DMIS), including a linearly increasing computational complexity to the problem horizon. In particular, the hierarchical problem structure is exploited to decompose the computationally intensive Hamiltonian minimization step into a set of low-dimensional optimizations. DMIS allows us to compute the direct fuel minimization problem including the vehicle and powertrain dynamics in a centralized manner to its full horizon while systematically tuning weighting factors that penalize passenger discomfort. For the first time, this study reveals that practically implemented sequential optimization exhibits similar fuel optimality as co-optimization when a certain level of passenger comfort is required.

Automation & Control Systems↗

Optimal ''image-based'' weighting for energy-resolved CT

This paper investigates a method of reconstructing images from energy-resolved CT data with negligible beam-hardening artifacts and improved contrast-to-nosie ratio (CNR) compared to conventional energy-weighting methods. Conceptually, the investigated method first reconstructs separate images from each energy bin. The final image is a linear combination of the energy-bin images, with the weights chosen to maximize the CNR in the final image. The optimal weight of a particular energy-bin image is derived to be proportional to the contrast-to-noise-variance ratio in that image. The investigated weighting method is referred to as ''image-based'' weighting, although, as will be described, the weights can be calculated and the energy-bin data combined prior to reconstruction. The performance of optimal image-based energy weighting with respect to CNR and beam-hardening artifacts was investigated through simulations and compared to that of energy integrating, photon counting, and previously studied optimal ''projection-based'' energy weighting. Two acquisitions were simulated: dedicated breast CT and a conventional thorax scan. The energy-resolving detector was simulated with five energy bins. Four methods of estimating the optimal weights were investigated, including task-specific and task-independent methods and methods that require a single reconstruction versus multiple reconstructions. Results demonstrated that optimal image-based weighting improved the CNR compared to energy-integrating weighting by factors of 1.15-1.6 depending on the task. Compared to photon-counting weighting, the CNR improvement ranged from 1.0 to 1.3. The CNR improvement factors were comparable to those of projection-based optimal energy weighting. The beam-hardening cupping artifact increased from 5.2% for energy-integrating weighting to 12.8% for optimal projection-based weighting, while optimal image-based weighting reduced the cupping to 0.6%. Overall, optimal image-based energy weighting provides images with negligible beam-hardening artifacts and improved CNR compared to energy-integrating and photon-counting methods.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

TU-EF-204-03: Task-Based KV and MAs Optimization for Radiation Dose Reduction in CT: From FBP to Statistical Model-Based Iterative Reconstruction (MBIR)

Purpose: The introduction of the highly nonlinear MBIR algorithm to clinical CT systems has made CNR an invalid metric for kV optimization. The purpose of this work was to develop a task-based framework to unify kV and mAs optimization for both FBP- and MBIR-based CT systems. Methods: The kV-mAs optimization was formulated as a constrained minimization problem: to select kV and mAs to minimize dose under the constraint of maintaining the detection performance as clinically prescribed. To experimentally solve this optimization problem, exhaustive measurements of detectability index (d’) for a hepatic lesion detection task were performed at 15 different mA levels and 4 kV levels using an anthropomorphic phantom. The measured d’ values were used to generate an iso-detectability map; similarly, dose levels recorded at different kV-mAs combinations were used to generate an iso-dose map. The iso-detectability map was overlaid on top of the iso-dose map so that for a prescribed detectability level d’, the optimal kV-mA can be determined from the crossing between the d’ contour and the dose contour that corresponds to the minimum dose. Results: Taking d’=16 as an example: the kV-mAs combinations on the measured iso-d’ line of MBIR are 80–150 (3.8), 100–140 (6.6), 120–150 (11.3), and 140–160 (17.2), where values in the parentheses are measured dose values. As a Result, the optimal kV was 80 and optimal mA was 150. In comparison, the optimal kV and mA for FBP were 100 and 500, which corresponded to a dose level of 24 mGy. Results of in vivo animal experiments were consistent with the phantom results. Conclusion: A new method to optimize kV and mAs selection has been developed. This method is applicable to both linear and nonlinear CT systems such as those using MBIR. Additional dose savings can be achieved by combining MBIR with this method. This work was partially supported by an NIH grant R01CA169331 and GE Healthcare. K. Li, D. Gomez-Cardona, M. G. Lubner: Nothing to disclose. P. J. Pickhardt: Co-founder, VirtuoCTC, LLC Stockholder, Cellectar Biosciences, Inc. G.-H. Chen: Research funded, GE Healthcare; Research funded, Siemens AX.

60 APPLIED LIFE SCIENCES↗

Robust Optimal Experimental Design of Infinite-Dimensional Bayesian Nonlinear Inverse Problems

Abstract. We consider robust optimal experimental design (ROED) for nonlinear Bayesian inverse problems governed by partial differential equations (PDEs). An optimal design is one that maximizes some utility quantifying the quality of the solution of an inverse problem. However, the optimal design is dependent on elements of the inverse problem such as the simulation model, the prior, or the measurement error model. ROED aims to produce an optimal design that is aware of the additional uncertainties encoded in the inverse problem and remains optimal even after variations in them. We follow a worst-case scenario approach to develop a new framework for robust optimal design of nonlinear Bayesian inverse problems. The proposed framework (a) is scalable and designed for infinite-dimensional Bayesian nonlinear inverse problems constrained by PDEs; (b) develops efficient approximations of the utility, namely the expected information gain; (c) employs eigenvalue sensitivity techniques to develop analytical forms and efficient evaluation methods of the gradient of the utility with respect to the uncertainties against which we wish to be robust; and (d) employs a probabilistic optimization paradigm that properly defines and efficiently solves the resulting combinatorial max-min optimization problem. The effectiveness of the proposed approach is illustrated for optimal sensor placement problem in an inverse problem governed by an elliptic PDE.

Chowdhary, Abhijit↗

Optimization Plugin Library

The Optimization Plugin library ("op") is a lightweight general optimization solver interface. The primary purpose of op is to simplify the process of integrating different optimization solvers (serial or parallel) with scalable parallel physics engines. By design it has several features that help make this a reality. The core abstraction interface was developed to encompass a large class of optimization problems in an optimizer-agnostic way. This enables us to describe the optimization problem once and then use a variety of supported "op" optimizers with ideally no code-changes. The abstraction interface is made up of lightweight wrappers that make it easy to integrate with existing simulation codes. This makes integration less intrusive and should minimize changes to existing physics codes. The "op" interface includes an assortment of utility methods that help specify parallel communication patterns as well as methods to convert from optimization-specific interfaces to the more general "op" interface. Lastly a dynamic library linking interface is provided to allow for use of proprietary optimization engines without explicit reference in the source code, along with standard shared library interfaces for opensource engines.

Jekel, CharlesF↗

Signatures of optimal codon usage in metabolic genes inform budding yeast ecology

Reverse ecology is the inference of ecological information from patterns of genomic variation. One rich, heretofore underutilized, source of ecologically relevant genomic information is codon optimality or adaptation. Bias toward codons that match the tRNA pool is robustly associated with high gene expression in diverse organisms, suggesting that codon optimization could be used in a reverse ecology framework to identify highly expressed, ecologically relevant genes. To test this hypothesis, we examined the relationship between optimal codon usage in the classic galactose metabolism ( GAL ) pathway and known ecological niches for 329 species of budding yeasts, a diverse subphylum of fungi. We find that optimal codon usage in the GAL pathway is positively correlated with quantitative growth on galactose, suggesting that GAL codon optimization reflects increased capacity to grow on galactose. Optimal codon usage in the GAL pathway is also positively correlated with human-associated ecological niches in yeasts of the CUG-Ser1 clade and with dairy-associated ecological niches in the family Saccharomycetaceae. For example, optimal codon usage of GAL genes is greater than 85% of all genes in the genome of the major human pathogen Candida albicans (CUG-Ser1 clade) and greater than 75% of genes in the genome of the dairy yeast Kluyveromyces lactis (family Saccharomycetaceae). We further find a correlation between optimization in the GAL actose pathway genes and several genes associated with nutrient sensing and metabolism. This work suggests that codon optimization harbors information about the metabolic ecology of microbial eukaryotes. This information may be particularly useful for studying fungal dark matter—species that have yet to be cultured in the lab or have only been identified by genomic material.

59 BASIC BIOLOGICAL SCIENCES↗

Learning to Optimize Variational Quantum Circuits to Solve Combinatorial Problems

Quantum computing is a computational paradigm with the potential to outperform classical methods for a variety of problems. Proposed recently, the Quantum Approximate Optimization Algorithm (QAOA) is considered as one of the leading candidates for demonstrating quantum advantage in the near term. QAOA is a variational hybrid quantum-classical algorithm for approximately solving combinatorial optimization problems. The quality of the solution obtained by QAOA for a given problem instance depends on the performance of the classical optimizer used to optimize the variational parameters. In this paper, we formulate the problem of finding optimal QAOA parameters as a learning task in which the knowledge gained from solving training instances can be leveraged to find high-quality solutions for unseen test instances. To this end, we develop two machine-learning-based approaches. Our first approach adopts a reinforcement learning (RL) framework to learn a policy network to optimize QAOA circuits. Our second approach adopts a kernel density estimation (KDE) technique to learn a generative model of optimal QAOA parameters. In both approaches, the training procedure is performed on small-sized problem instances that can be simulated on a classical computer; yet the learned RL policy and the generative model can be used to efficiently solve larger problems. Furthermore, extensive simulations using the IBM Qiskit Aer quantum circuit simulator demonstrate that our proposed RL- and KDE-based approaches reduce the optimality gap by factors up to 30.15 when compared with other commonly used off-the-shelf optimizers.

97 MATHEMATICS AND COMPUTING↗

Pressurized Water Reactor Gadolinia Pin Location Optimization

This report presents the results of lattice optimization studies performed to find optimum locations for gadolinia burnable absorber (BA) rods in pressurized water reactor (PWR) lattice fuel designs. Initial excess reactivity suppression allows core designers to further improve operational economics by extending cycle length. Gadolinia BAs are commonly used in boiling water reactor assembly designs for this purpose. In recent years, gadolinia absorbers have been used in PWR designs owing to their longer effectiveness for reactivity suppression compared with common BAs used in PWR assemblies. This report examines the optimum gadolinia pin placement in 17 × 17 PWR lattices at different fuel and gadolinia concentrations for optimized lattice performance, using the SCALE/Polaris lattice physics code. The completed work is continuation of the Light Water Reactor LEU+ Lattice Optimization (ORNL/TM-2021/2366) project. An optimization driver called the metaheuristic optimization tool (MOT) is used to automate domain space exploration and optimization of the lattice designs. Heuristics from previous light-water reactor (LWR) lattice optimization studies were used to construct the objective function and define the domain space for optimization. This work successfully demonstrated that the optimization algorithms of MOT can generate feasible, nonproprietary PWR lattice designs with gadolinia.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Intrusive Uncertainty Quantification and Optimal Experiment Design in the Open-Source Pyomo Ecosystem

This contribution describes ParmEst and Pyomo.DoE, two pillars of the open-source Python-based Pyomo ecosystem for computational optimization with (partial differential) algebraic equation mathematical models. Specifically, ParmEst facilitates intrusive frequentist parameter estimation (PE) and uncertainty quantification (UQ) through built-in features, such as covariance matrix estimation, bootstrapping, and likelihood ratio tests. Complementary, Pyomo.DoE enables optimal experiment design by maximizing various metrics of the Fisher information matrix, such as A-optimality (trace), D-optimality (determinant), E-optimality (minimum eigenvalue), and ME-optimality (condition number). ParmEst and Pyomo.DoE can solve high-dimensional optimization problems by leveraging the model structure and exact derivative information. Finally, we will discuss future opportunities to integrate PE and UQ capabilities with optimization under uncertainty, including robust optimization with non-convex models via PyROS.

97 MATHEMATICS AND COMPUTING↗

Modeling and Optimization of a Nuclear Integrated Energy System for the Remote Microgrid on El Hierro

Nuclear microreactors are a potential technology to provide heat and electricity for remote microgrids. There is potential for the microgrid on the island of El Hierro to use a microreactor, within an integrated energy system (IES), to generate electricity and provide desalinated water. This work proposes a workflow for optimizing and analyzing IESs for microgrids. In this study, an IES incorporating a microreactor, thermal energy storage (TES) system, combined heat and power plant, and a thermal desalination plant was designed, optimized, and analyzed using Idaho National Laboratory’s Framework for Optimization of Resources and Economics (FORCE) toolset. The optimization tool, Holistic Energy Resource Optimization Network (HERON), was used to determine the optimal capacity sizes and dispatch for the reactor and thermal energy storage systems to meet demand. The optimized reactor and TES sizes were found to be 11.61 MWth and 58.47 MWhth, respectively, when optimizing the IES to replace 95% of the island’s existing diesel generation needs. A dynamic model of the system was created in the Modelica language, using models from the HYBRID repository, to analyze and verify the dispatch from the optimizer. The dynamic model was able to meet the ramp rates while maintaining reactor power with minimal control adjustments.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Coupled wind turbine design and layout optimization with nonhomogeneous wind turbines

Abstract. In this study, wind farms were optimized to show the benefit of coupling complete turbine design and layout optimization as well as including two different turbine designs in a fixed 1-to-1 ratio in a single wind farm. For our purposes, the variables in each turbine optimization include hub height, rotor diameter, rated power, tower diameter, tower shell thickness, and implicit blade chord-and-twist distributions. A 32-turbine wind farm and a 60-turbine wind farm were both considered, as well as a variety of turbine spacings and wind shear exponents. Structural constraints as well as turbine costs were considered in the optimization. Results indicate that coupled turbine design and layout optimization is superior to sequentially optimizing turbine design, then turbine layout. Coupled optimization results in an additional 2 %–5 % reduction in the cost of energy compared to optimizing sequentially for wind farms with turbine spacings of 8.5–11 rotor diameters. Smaller wind farms benefit even more from coupled optimization. Furthermore, wind farms with closely spaced wind turbines can greatly benefit from nonuniform turbine design throughout the farm. Some of these wind farms with heterogeneous turbine design have an additional 10 % cost-of-energy reduction compared to wind farms with identical turbines throughout the farm.

Stanley, Andrew P. J.↗

Roofline Analysis in the Intel® Advisor to Deliver Optimized Performance for applications on Intel® Xeon Phi™ Processor

In this session we show, in two case studies, how the roofline feature of Intel Advisor has been utilized to optimize the performance of kernels of the XGC1 and PICSAR codes in preparation for Intel Knights Landing architecture. The impact of the implemented optimizations and the benefits of using the automatic roofline feature of Intel Advisor to study performance of large applications will be presented. This demonstrates an effective optimization strategy that has enabled these science applications to achieve up to 4.6 times speed-up and prepare for future exascale architectures. # Goal/Relevance of Session The roofline model [1,2] is a powerful tool for analyzing the performance of applications with respect to the theoretical peak achievable on a given computer architecture. It allows one to graphically represent the performance of an application in terms of operational intensity, i.e. the ratio of flops performed and bytes moved from memory in order to guide optimization efforts. Given the scale and complexity of modern science applications, it can often be a tedious task for the user to perform the analysis on the level of functions or loops to identify where performance gains can be made. With new Intel tools, it is now possible to automate this task, as well as base the estimates of peak performance on measurements rather than vendor specifications. The goal of this session is to demonstrate how the roofline feature of Intel Advisor can be used to balance memory vs. computation related optimization efforts and effectively identify performance bottlenecks. A series of typical optimization techniques: cache blocking, structure refactoring, data alignment, and vectorization illustrated by the kernel cases will be addressed. # Description of the codes ## XGC1 The XGC1 code [3] is a magnetic fusion Particle-In-Cell code that uses an unstructured mesh for its Poisson solver that allows it to accurately resolve the edge plasma of a magnetic fusion device. After recent optimizations to its collision kernel [4], most of the computing time is spent in the electron push (pushe) kernel, where these optimization efforts have been focused. The kernel code scaled well with MPI+OpenMP but had almost no automatic compiler vectorization, in part due to indirect memory addresses and in part due to low trip counts of low-level loops that would be candidates for vectorization. Particle blocking and sorting have been implemented to increase trip counts of low-level loops and improve memory locality, and OpenMP directives have been added to vectorize compute-intensive loops that were identified by Advisor. The optimizations have improved the performance of the pushe kernel 2x on Haswell processors and 1.7x on KNL. The KNL node-for-node performance has been brought to within 30% of a NERSC Cori phase I Haswell node and we expect to bridge this gap by reducing the memory footprint of compute intensive routines to improve cache reuse. ## PICSAR is a Fortran/Python high-performance Particle-In-Cell library targeting at MIC architectures first designed to be coupled with the PIC code WARP for the simulation of laser-matter interaction and particle accelerators. PICSAR also contains a FORTRAN stand-alone kernel for performance studies and benchmarks. A MPI domain decomposition is used between NUMA domains and a tile decomposition (cache-blocking) handled by OpenMP has been added for shared-memory parallelism and better cache management. The so-called current deposition and field gathering steps that compose the PIC time loop constitute major hotspots that have been rewritten to enable more efficient vectorization. Particle communications between tiles and MPI domain has been merged and parallelized. All considered, these improvements provide speedups of 3.1 for order 1 and 4.6 for order 3 interpolation shape factors on KNL configured in SNC4 quadrant flat mode. Performance is similar between a node of cori phase 1 and KNL at order 1 and better on KNL by a factor 1.6 at order 3 with the considered test case (homogeneous thermal plasma).

97 MATHEMATICS AND COMPUTING↗

Multi-scale and multi-physics analysis, design optimization, and experimental validation of heat exchangers utilizing high performance, non-round tubes

Air-to-refrigerant heat exchangers (HXs) are fundamental components in HVAC&R systems, and considerable research has been dedicated designing continually smaller, lighter, and more efficient HX designs. In recent years, researchers have leveraged advancements in Computational Fluid Dynamics (CFD), Finite Element Analysis (FEA), and optimization algorithms to consider primary tube shape and topology optimization to design highly compact, high performance HXs for a multitude of applications. In this research, we present a computationally efficient, comprehensive, multi-scale, and multi-physics analysis and optimization method for air-to-refrigerant HXs featuring automated CFD and FEA simulations and approximation-assisted optimization. This methodology was utilized to design HXs with shape-optimized, non-round tubes which outperform current state-of-the-art tube-fin HXs without compromising structural integrity. The optimal HXs were shown to deliver similar thermal performance to the baseline HXs while also achieving more than 20% reductions in airside pressure drop and core envelope volume and more than 30% reduction in internal volume. Comprehensive experimental validation of the optimization methodology was conducted through the testing of two prototypes in a standardized wind tunnel facility under multiple operating conditions. For prototype #1 under dry evaporator conditions, the predicted heat load agreed within ± 10% of measured values and the predicted airside pressure drop agreed within ± 30%, while for dehumidifying conditions, the predicted sensible and latent heat loads agreed within ± 10% and ± 20% of the measured values, respectively. For prototype #2, the predicted condenser heat load agreed within ± 3.0% of measured values, and the predicted airside pressure drop agreed within ± 27%. The acceptable agreement between simulation and experimental results for the present application highlights the flexibility of the novel optimization methodology to design next generation HXs with improved performance and reduced volume, weight, and environmental impact.

42 ENGINEERING↗

Variable-fidelity multipoint aerodynamic shape optimization with output-based adapted meshes

This work presents a method to control the discretization error in multipoint aerodynamic shape optimization using output-based adapted meshes. The meshes are adapted via adjoint-based error estimates, taking into account both the objective and constraint output errors. A multi-fidelity optimization framework is then developed by taking advantage of the variable fidelity offered by adaptive meshes. The objective functional and its sensitivity at each design point (operating condition) are first evaluated on the same initial coarse mesh, which is then subsequently adapted for each design point individually as the shape optimization proceeds. The effort to set up the optimization is minimal since the initial mesh can be fairly coarse and easy to generate. As the shape approaches the optimal design, the mesh at each design point becomes finer, in regions necessary for that particular operating condition. The multi-fidelity framework is tightly coupled with the objective error estimation to ensure the optimization accuracy at each fidelity. Computational savings arise from a reduction of the mesh size when the design is far from optimal and avoiding an exhaustive search on low-fidelity meshes. The proposed method is demonstrated on multipoint drag minimization problems of a transonic airfoil with lift and area constraints. Improved accuracy and efficiency are shown compared to traditional fixed-fidelity optimization with a fixed computational mesh.

42 ENGINEERING↗

Development of novel dynamic machine learning-based optimization of a coal-fired power plant

The increasing fraction of intermittent renewable energy in the electrical grid is resulting in coal-fired boilers now routinely ramp up and down. The current state-of-the-art operation for such boilers is to apply steady-state, neural network-based optimization to make control decisions in real-time, and this report demonstrates the feasibility of extending this to dynamic, neural network-based optimization using a long short-term memory neural network. A simplified numerical simulation of a t-fired coal boiler and supporting equipment is used to represent a real plant subjected to both steady-state, neural network-based optimization and dynamic, neural network-based optimization. Using the same intervals and a particle swarm optimization algorithm, the dynamic optimization outperforms the steady-state optimization and realizes up to 4.58% improvement in thermal efficiency. Dynamic optimization with a long short-term memory neural network is shown to both be feasible and beneficial for operation of a coal-fired boiler under changing load.

42 ENGINEERING↗